MHC class II genes mediate susceptibility and resistance to coronavirus infections in bats

Understanding the immunogenetic basis of coronavirus (CoV) susceptibility in major pathogen reservoirs, such as bats, is central to inferring their zoonotic potential. Members of the cryptic Hipposideros bat species complex differ in CoV susceptibility, but the underlying mechanisms remain unclear. The genes of the major histocompatibility complex (MHC) are the best understood genetic basis of pathogen resistance, and differences in MHC diversity are one possible reason for asymmetrical infection patterns among closely related species. Here, we aimed to link asymmetries in observed CoV (CoV‐229E, CoV‐2B and CoV‐2Bbasal) susceptibility to immunogenetic differences amongst four Hipposideros bat species. From the 2072 bats assigned to their respective species using the mtDNA cytochrome b gene, members of the most numerous and ubiquitous species, Hipposideros caffer D, were most infected with CoV‐229E and SARS‐related CoV‐2B. Using a subset of 569 bats, we determined that much of the existent allelic and functional (i.e. supertype) MHC DRB class II diversity originated from common ancestry. One MHC supertype shared amongst all species, ST12, was consistently linked to susceptibility with CoV‐229E, which is closely related to the common cold agent HCoV‐229E, and infected bats and those carrying ST12 had a lower body condition. The same MHC supertype was connected to resistance to CoV‐2B, and bats with ST12 were less likely be co‐infected with CoV‐229E and CoV‐2B. Our work suggests a role of immunogenetics in determining CoV susceptibility in bats. We advocate for the preservation of functional genetic and species diversity in reservoirs as a means of mitigating the risk of disease spillover.


| INTRODUC TI ON
Bats are the most notorious vectors and zoonotic origin for many coronaviruses (CoVs) (Banerjee et al., 2019;Corman et al., 2018;Ruiz-Aravena et al., 2021), including both SARS-CoVs (Li et al., 2005;Zhou et al., 2020), MERS-CoV (Ithete et al., 2013) and two common cold agents, HCoV-229E (Corman et al., 2015) and HCoV-NL63 (Huynh et al., 2012). In fact, CoV sequence variants make up more than a third of all known viral infections in bats (Letko et al., 2020), and more speciose bat communities harbour more diverse CoV strains (Anthony et al., 2017). In humans, host genetics substantially impact transmission susceptibility, disease severity and mortality risk from SARS-CoV-2 (Pathak et al., 2021) and other CoVs (Ovsyannikova et al., 2020). In addition to receptors, such as the transmembrane serine protease 2 or angiotensin converting enzyme 2, which are frequently involved in cell entry of CoVs (Hoffmann et al., 2020), several alleles of the human equivalent of the major histocompatibility complex (MHC)-the human leukocyte antigen (HLA)-are putatively connected to adaptive immune reactions against infections with SARS-CoV-2 (Migliorini et al., 2021) and other CoVs (Chu et al., 2014;Loyal et al., 2021). However, in contrast with humans, very little is known about the relationship between the MHC and resistance to CoVs in their natural reservoirs.
The multigene cluster of the MHC is the best characterised genetic basis of pathogen resistance in jawed vertebrates (Kaufman, 2018a;Sommer, 2005). The genomic region encodes a suite of structurally related yet distinct cell surface glycoproteins-MHC class I and II-which present self-and non-self-peptides in their peptide-binding regions to CD8 + and CD4 + T cells. Foreign peptides then trigger pathogen-specific immune responses. MHC class I and II molecules present peptides derived from intracellular or extracellular pathogens, respectively (Neefjes et al., 2011). Yet, intracellular peptides may be degraded via autophagy and presented by MHC class II molecules instead (Dengjel et al., 2005;Harman et al., 2009;Laing et al., 2019). Aside from cross-presentation, host control over commensal bacteria, which modulate host immunity (Fleischer et al., 2022;Han et al., 2022;Kubinak et al., 2015), is mediated via the MHC class II and associated T-cell responses (Cao et al., 2012;Roland et al., 2020).
Several mutually non-exclusive hypotheses were put forward as drivers of the coevolutionary arms race between host and pathogen and to explain the exceptional polymorphism observed at the MHC (Radwan et al., 2020;Spurgin & Richardson, 2010). At their core, each hypothesis explains how the identity, diversity and frequency of MHC alleles and supertypes (STs; i.e. MHC alleles grouped together based on shared antigen binding properties, Doytchinova et al., 2004;Lighten et al., 2017) relate to host resistance: individuals with functionally heterogeneous MHC alleles/STs, for instance, are able to fend off a greater variety of pathogens (called heterozygote/ divergent allele advantage; Lenz et al., 2009;Takahata & Nei, 1990;Wegner et al., 2003). Since pathogens evolve to evade common MHC alleles/STs, advantageous (often rare) MHC alleles/STs may rapidly increase in frequency in the host population, resulting in cyclic negative frequency-dependent selection (Eizaguirre et al., 2012b;Meyer-Lucht & Sommer, 2005;Phillips et al., 2018), and distinct pathogen communities result in fluctuating selection across space and time (Eizaguirre et al., 2012a;Hill et al., 1991). The same mechanisms seem to govern resistance to CoVs in humans: HLA (MHC class I and II) variants are linked to COVID-19 symptoms and disease severity (Bruchez et al., 2020;Castelli et al., 2021;Vietzen et al., 2021), the frequency of two HLA haplotypes correlates negatively and positively with country-wide infection gradients in Italy (Pisanti et al., 2020), and HLA-SARS-CoV-2 associations vary globally (Migliorini et al., 2021).
Bat immunity is frequently treated as a black box owing to its peculiarities . For example, bats employ antibodies in defence against CoV infections (Müller et al., 2007), but their immune system seems rather to tolerate than purge viral infections (Banerjee et al., 2020), often leading to prolonged infectious periods, despite appearing physically healthy (Irving et al., 2021;Munster et al., 2016;Subudhi et al., 2017). In addition, bats are highly gregarious and often share roosting caves with different, sometimes phylogenetically closely related bat species, further facilitating cross-species viral spillover events and sustaining shared pathogen-mediated selection. Pathogen-mediated selection is likely at the root of the exceptional allelic and functional diversity of the MHC class I and II region observed in bats (e.g. Qurkhuli et al., 2019;Salmier et al., 2016;Schad et al., 2011). Yet, few studies to date draw connections between any of the many viruses, let alone CoVs, infecting bats and host MHC genetics (e.g. Astroviridae-MHC class II (Fleischer et al., 2022); Influenza A viruses-MHC class II (Giotis et al., 2019;Karakus et al., 2019)). Moreover, the consequence of co-habitation of bats with shared functional MHC diversity has never been investigated, even though it represents an essential step towards understanding CoV prevalence and persistence in natural host populations, and its potential risk to public health.
Members of the particularly speciose, palaeotropical leafnosed bat family Hipposideridae are host to a number of corona-, Ebola-und paramyxoviruses with zoonotic potential (Annan et al., 2013;Drexler et al., 2012;Hayman et al., 2012;Pfefferle et al., 2009;Suu-Ire et al., 2022). Sub-Saharan hipposiderids harbour an ancestral form of the common cold agent, the alpha-CoV-229E, and two SARS-related beta-CoVs (Corman et al., 2015;Pfefferle et al., 2009). Recent molecular, ecological and behavioural evidence was used to differentiate three morphologically indistinguishable species of the Hipposideros caffer complex (Baldwin et al., 2014Vallo et al., 2008), and it was suspected that the species also differ in CoV susceptibility (Baldwin, 2015). In the present study, we aim to understand how MHC class II genes mediate susceptibility and resistance to CoV infections in hipposiderids. Whereas MHC class I traditionally is associated with an anti-viral immune cascade, cross-presentation of, for example, autophagically derived viral peptides at MHC class II molecules might be an important anti-viral mechanism in bats (Laing et al., 2019). Additionally, CoVs replicate enterically in bats . Resistance to enterically replicating viruses was previously found to be linked indirectly to MHC class II via host microbiota (Fleischer et al., 2022). We therefore hypothesise if differences in infection prevalence exist among members of the Hipposideros caffer complex, infection likelihood might be associated with allelic and functional diversity of the MHC DRB class II region. Our study found differences in infection likelihood between species and identifies consistent associations between CoV infection likelihood and functional MHC supertypes.  (2015) and Meyer et al. (in review). If possible, all bats were classified to species level using morphological characteristics. Morphometric details such as forearm length or weight were described elsewhere . Two minimally invasive wing punches (3 mm) were taken from each bat and stored in molecular-grade ethanol at −20°C for The extraction followed an ammonium acetate protocol (Nicholls et al., 2000) with slight modifications from T. Halczok, University of Greifswald. Building on previous primer designs for the Hipposideros species complex (Vallo et al., 2008), the mtDNA cytochrome b gene (cytb) was amplified by polymerase chain reaction (PCR) using adapted primers suitable for high-throughput Illumina sequencing.

| Sample collection
After sequencing, the cytb gene was confirmed by homology analysis using the NCBI BLAST search. Subsequently, all sequences were analysed in Geneious 11.1.5 (https://www.genei ous.com) and assigned to the lineages B, C or D of the H. caffer species complex (henceforth called species, Baldwin et al., 2014Baldwin et al., , 2021Vallo et al., 2008) or the sympatric species H. abae using the MAFFT alignment tool (Katoh & Standley, 2013; for more detail see Appendix S1; Table S1).

| Virus screening: RNA purification and CoV characterisation
RNA was purified from approximately 20 mg of faecal material suspended in 500 μL RNAlater stabilising solution using the MagNA Pure 96 system (Roche) with elution volumes set at 100 μL. We used a real-time reverse transcription-PCR assay designed to detect several alpha-and beta-CoVs and genetically related bat CoVs using the SSIII RT-PCR kit (Life Technologies) and a cycling protocol in a LightCycler 480 (Roche) as described previously (Corman et al., 2015;Drexler et al., 2009;Pfefferle et al., 2009; see Appendix S1; Table S2). Bats were categorised as positive for a specific CoV if the CT value was equal or smaller than 38.0 (Pfefferle et al., 2009).

| Sequencing of MHC class II DRB exon 2 loci
A 171 bp fragment within the MHC class II DRB exon 2 loci of 575 Hipposideros samples was amplified using primers modified from Schad et al. (2011) and quality checked in Geneious. The JSN primers are widely used for chiropteran MHC class II DRB as they have an exceptional ability to amplify the region in a variety of species. Hipposideros samples were MHC genotyped using an Illumina F I G U R E 1 (a) Community composition of Hipposideros bat species captured at five cave locations in Ghana and (b) number of individuals uninfected, single or co-infected with several alpha-and beta-CoVs.
platform (see Appendix S1). The samples covered all locations, lineages and CoV infections (Table S3 and S4) and, collectively, reached beyond the threshold sample size (>200) suggested for wild populations (Gaigher et al., 2019). We ran duplicates for 53 samples to assess repeatability.

| MHC class II allele characterisation using ACACIA
The MHC class II DRB exon 2 sequences were analysed using the genotyping pipeline ACACIA (Allele CAlling proCedure for Illumina Amplicon sequencing data; ; https://gitlab.com/psc_santo s/ACACIA), most recently employed to analyse sequences from a neotropical bat species (Fleischer et al., 2022). ACACIA first filters sequence quality, before confirming MHC DRB sequences by homology analysis using NCBI BLAST search based on other chiropteran MHC DRB sequences.
Forward and reverse reads were merged with a minimum overlap of 50 base pairs (bp) and a maximum overlap of 250 bp. Quality filtering removed sequences with a Phred quality score value <30 and a q-value <90. For alleles to be retained, the minimum number of reads per allele was set to 10, and the lowest percentage of the total reads of an individual was 1%. The ACACIA workflow and post hoc elimination of singletons and alleles with low reliability preserved reliable MHC allele information for a total of 569 bats (Appendix S1; Figure S1 and Table S3).

| MHC supertype assignment
Allelic MHC diversity was then grouped into functional supertypes based on shared amino acid motifs at positively selected sites (PSSs), following the assumption that PSSs likely belong to, or are closely linked to, functionally important antigen-binding sites (Roved et al., 2022;Schwensow et al., 2019;Sepil et al., 2013).
PSSs were identified for each species separately using CODEML integrated in the program PAML4 (Phylogenetic Analysis by Maximum Likelihood; Yang, 2007) and run in the graphical interface PAML-X GUI (Xu & Yang, 2013). Four PAML models were tested: M1a (nearly neutral), M2a (positive selection), M7 (beta) and M8 (beta and ω). M2a and M8 performed equally well, evidencing selection acting on specific sites. For subsequent supertype assignment across species, a total of 14 'consensus' PSS sites were selected based on being (a) identified by both M2a and M8 with a posterior probability of at least 95% and (b) selected in at least 3 of the 4 species ( Figure S2; Table S5).
Antigen-binding specificity can be quantified by z-values describing the physiochemical properties of amino acids encoded by the codon present at PSSs (Sandberg et al., 1998). A matrix containing the z values of each allele's PSS amino acid was used in the functions find.clusters and dapc (i.e. discriminant analysis of principal components) of the 'adegenet' R package (Jombart et al., 2010) to cluster alleles into groups (i.e. MHC supertypes or STs) with similar binding functionality (see Appendix S1, Figure S3).

| Community composition and CoV infection patterns
Differences in community and abundance between the 2072 sampled hipposiderids were assessed using a permutational analysis of variance (PERMANOVA). A x 2 -test was run to compare proportional differences in single and co-infections with any of the three CoVs between species.

| MHC diversity
Individual MHC allelic and ST diversity were compared among roosting locations, species and their interaction using generalised linear models (GLMs) with Poisson distribution. To assess whether MHC allelic and supertype diversity present was captured with the sampling effort across locations and for each species, we used the specaccum function, and, to confirm whether an asymptote was reached, the specpool function of the 'vegan' r package.

| Evidence for balancing selection and/or parallel evolution
Owing to common ancestry, sites under positive selection should be characterised by fewer synonymous codons if balancing selection maintained trans-species polymorphism (TSP) rather than convergence based on independent parallel evolution (Lundberg et al., 1992).
To differentiate between these mechanisms, we compared the proportion of identical codons encoding the same amino acid at PSSs between species pairs (Lenz et al., 2013).

| Associations between CoV infection and MHC diversity or supertype identity
We identified associations between the presence of an MHC ST and CoV infections using the probabilistic model of co-occurrence (Veech, 2013) programmed in the 'cooccur' r package (Griffith et al., 2016). A positive association between an MHC ST and CoV infection is assumed when the observed co-occurrence is significantly higher than the expected co-occurrence and vice versa. In other words, a higher co-occurrence implies higher probability of individuals with the particular MHC ST harbouring an infection and, thus, suggests susceptibility. By contrast, a significantly lower cooccurrence suggests resistance. The co-occurrence model was run separately for each species due to differences in MHC ST profiles, but concurrently for all CoV infections present in each species. We only included MHC STs represented in at least 10 individuals to guarantee a reasonable sample size.
The results from the co-occurrence analysis were confirmed with generalised linear mixed effects models (GLMMs) using CoV infections as binomial response variable, and number of STs and its quadratic term as well as the specific STs identified by the co-occurrence analyses as explanatory variables. Number of STs is a proxy for functional MHC diversity given its correlative link to other measures of MHC diversity (Appendix S1; Figure S4).
Capture period was set as random effect. Final models retained only explanatory variables suggested by an information theoretic approach using the dredge function from the 'MuMIn' r package (Bartoń, 2009). False discovery rate correction was applied when co-occurrence analysis had identified multiple STs and, thus, multiple GLMMs were run on the same response variable (Benjamini & Hochberg, 1995). The alpha-CoV-229E and the beta-CoV-2B infected all Hipposideros species, whereas the more basal beta-CoV-2Bbasal was only found in H. abae and H. caffer D (Figure 1b). The prevalence of CoVs differed between Hipposideros species (prevalence:

| MHC allelic and supertype diversity in the Hipposideros species complex
We detected a total of 165 MHC class II alleles from the 569 Hipposideros bats sampled across five different locations in Sub-Saharan Ghana (repeatability: 94.3%). The 171 bp long fragments mapped inside the DRB exon 2 and 8 PSSs were congruent with antigen binding sites of the human HLA-DRB*0101, while the others were within one amino acid position to human antigen binding sites (Table S5). With the exception of W61, conserved sites in the hipposiderid sequences coincided with conserved sites of the human reference sequence (Brown et al., 1993), which is consistent with MHC class II DRB exon 2 sequence variation previously found in bats (Schad et al., 2011).  (Table S6a) and between 3.80 (±1.14 SD) and 5.25 (±2.09 SD) among Hipposideros species (Table S6b). H. caffer D in the Forikrom cave showed exceptionally high allelic diversity (mean 6.97 ± 2.13 SD; Table S6c). Overall, we likely captured the present allelic diversity of the four species ( Figure S6A,B) and reach levels of allelic diversity similar to that of other Chiroptera (Table S7).
Functional diversity was also similar between locations, species and for each species per location (Table S6a-

| Codon usage analysis
We found identical amino acid sequences for a total of 34 of the 165 MHC class IIB DRB exon 2 alleles ( Table 1). Three alleles that grouped within ST12 were shared among all species and another three also within ST12 were shared among all lineages from the H. caffer species complex. This encourages the argument for functional redundancy among MHC amino acid alleles, and suggests balancing selection to maintain the observed TSP.
Following these results, we estimated the frequency of nonsynonymous to synonymous codons used at PSSs in the hipposiderids to formally test whether the pattern was more likely to arise from parallel evolution or balancing selection. We found an average codon similarity of 92.38% (±0.52 SD) between species (Table 1). Hence, fewer synonymous codon changes were observed than expected from simulated data (all p-values < .001; Figure S7).

| CoV-MHC diversity and supertype associations
The co-occurrence model identified and subsequent GLMMs confirmed a total of four MHC STs to be associated with CoV-229E infections (Figure 2b, all p-values and effect sizes are reported in Table S8; GLMM results are reported in Table S9) in H. caffer C were positively associated with CoV-2B.

| Follow-up: Co-infections, quantitative resistance/susceptibility and host fitness
Motivated by our findings of consistently pleiotropic effects of MHC ST12, which is shared among all species and contains highly conserved alleles, three hypotheses emerged. First, if the ST is indeed functionally divergent, one might expect fewer individuals with CoV-229E-CoV-2B co-infections among those carrying MHC ST12. To test this, we used two x 2 -Test of Independence comparing the number of uninfected, singly infected and co-infected individuals with or without MHC ST12 and visualised the results using the assoc function of the 'vcd' r package (Zeileis et al., 2007).
As hypothesised, significantly fewer individuals carrying MHC ST12 were co-infected with CoV-229E and CoV-2B (Figure 3a, Table S10). F I G U R E 2 (a) Major histocompatibility complex (MHC) supertype frequencies and (b) associations with three common CoV-infections in four Hipposideridae species from Ghana. Dashed line marks the minimum ST frequency (n = 10) at which ST was still included in the co-occurrence analysis. The significance of the co-occurrence analysis is presented as: * < .05; ** < .01; *** < .001. 'effects' represent standardized effect sizes and indicate positive (implying susceptibility) and negative (implying resistance) association between MHC ST and CoV infection in pink and green, respectively. Brackets around asterisk imply that the results were not supported by the confirmatory GLMMs. The best model showed a higher body condition for members of

H. abae and H. caffer B, but a lower body condition in bats infected
with CoV-229E or carrying MHC ST12 (Figure 4; Table S11). The CoV-2B infection status, number of STs and the interaction were not retained in the top models (Table S12).

| DISCUSS ION
Bats are hosts, reservoirs and vectors for a variety of viruses and the principal source of CoVs with zoonotic potential (Anthony et al., 2017;Letko et al., 2020;Ruiz-Aravena et al., 2021). Yet, few studies aimed to understand the genetic basis of virus resistance and susceptibility in bats (e.g. Fleischer et al., 2022), and none in relation to CoVs. Here we report asymmetrical CoV infection patterns and shared allelic and functional MHC class II diversity among four Hipposideros bat species co-inhabiting several roosting caves in Ghana. We found several positive and negative associations be-

F I G U R E 4
Body condition between uninfected and CoV-229E-infected bats and its variation according to whether major histocompatibility complex (MHC) ST12 is part of an individual's MHC constitution.
Globally, hipposiderids and their sister family Rhinolophidae are considered major CoV reservoirs (Annan et al., 2013;Corman et al., 2015;Gouilh et al., 2011;Quan et al., 2010). Speciose assemblages also often show high rates of CoV evolution, suggesting links between bat and CoV diversity (Anthony et al., 2017). At the same time, CoV infection prevalence declines in more diverse roosting communities in Ghana (Meyer et al., in review), supporting the idea that more diverse host assemblages alter transmission dynamics (Civitello et al., 2015). In such systems, cross-species spillover events are likely frequent (Latinne et al., 2020), and CoVs may adapt rapidly and extensively (e.g. <10%-40% amino acid divergence in spike protein) to novel host (Boni et al., 2020;Crossley et al., 2012;Lau et al., 2010Lau et al., , 2012. This raises the possibility that episodic diversifying selection in CoV strains facilitates spillover between animal hosts, similar to the mechanism driving variant evolution in SARS-CoV-2 (Tay et al., 2022). If competent CoV-hosts dominate a diverse species community, novel CoV strains may readily evolve and possibly switch host to become a wildlife and public health concern.
The allelic MHC class II diversity of four hipposiderid bats was modest in comparison with other Chiroptera. Though the 79 alleles characterised in H. caffer C alone rallies diversity typically only found in birds (Gillingham et al., 2016;Whittingham et al., 2018) and few exceptionally diverse bats with large geographical ranges, such as Carollia perspicillata (Qurkhuli et al., 2019). Interestingly, H. caffer C's geographical range is much smaller than that of C. perspicillata but instead overlaps with several other hipposiderids (Vallo et al., 2008) and co-habitation is common (Baldwin, 2015). Sociality, with its many opportunities for cross-species transmission of shared pathogens, rather than migration and its associated need to adapt to geographically distinct pathogens, might thus select for high MHC diversity, as also suggested for some colony forming birds (Minias et al., 2017) (Li et al., 2021;Zhao et al., 2014). In terms of functional MHC class II diversity, hipposiderids shared between 8 and 11 of the 12 identified STs, which is comparable with the number of STs detected in Artibeus jamaicensis (Fleischer et al., 2022).
Hipposiderids also shared several MHC alleles. Such high degree of retained functional and allelic diversity in the estimated 5.5-4.4 million years since their divergence (Baldwin, 2015) could be an argument for balancing selection driven by common pathogen pressure and frequent cross-species transmission in shared roosting caves. However, introgression of MHC alleles via hybridisation of sympatric species could be a confounding factor (Nadachowska-Brzyska et al., 2012). Balancing selection and hybridisation, for example, explain MHC diversity among Eurasian newts (Dudek et al., 2019;Gaczorek et al., 2022) and green lizards (Sagonas et al., 2018) and suggest an adaptive role of introgression in maintaining genetic diversity in habitats with shared selection pressure (Marques et al., 2019).
Several MHC class II STs were positively or negatively associated with CoV-229E, CoV-2B and, to a lesser degree, with the more ancestral CoV-2Bbasal strain. The associations were most pronounced in the heavily infected H. caffer D. Despite the notion that MHC class I plays a more significant role in viral resistance than MHC class II (Kaufman, 2018b), shared immunological pathways (e.g. MHC class II transactivator (CIITA) induces presentation of MHC class I molecules (Bruchez et al., 2020)) and antigen cross-presentation (Neefjes et al., 2011) suggest coordinated anti-pathogen responses. In fact, antigens presented by MHC class II molecules can itself trigger antiviral immune responses. When antigen-presenting cells detect intracellular pathogens and present antigens, the Th1 group of CD4 + T cells become activated and stimulate cytotoxic T cells, macrophages and anti-viral antibody production by B cells. Additionally, CoVs replicate in the enteric tract of many wildlife species, including bats . At the mucosal interface such as the intestinal lumen, Th17 cells produce the anti-microbial peptide generating cytokine interleukin 22 (Liang et al., 2006), and coordinate the secretion of IgA antibodies by B cells (Cao et al., 2012). Both, but IgAs, in particular, play important roles in the cross-talk between commensal gut bacteria and host immunity (Huus et al., 2021;Roland et al., 2020), and several studies have confirmed a link between gut microbial diversity and host MHC diversity and constitution (Bolnick et al., 2014;Davies et al., 2022;Silverman et al., 2017). This means competition and production of immunomodulatory molecules by MHC-class II-regulated commensals might feasibly aid anti-viral responses in bats, and was previously proposed to explain infection patterns with an enterically replicating Astrovirus in a neotropical bat (Fleischer et al., 2022) and Adenovirus-infected lemurs . Another possibility is that certain viruses exploit MHC class II receptor proteins for cell entry (Giotis et al., 2019;Karakus et al., 2019). Regardless, our study is not the first to detect associations between specific MHC alleles/STs and viral infections (e.g. H1N1 influenza (Luckey et al., 2019) (Milinski et al., 2005); birds (Buchholz et al., 2004); mammals (Schwensow et al., 2008) The diversity of MHC and receptor proteins, such as angiotensinconverting enzyme 2, are candidate host traits most informative to evaluate cross-species transmission potential (Frank et al., 2022), and likely explain mechanistically how greater host (genetic) diversity keeps pathogen prevalence in check (Civitello et al., 2015).
Although spillover risks likely depend on a variety of ecological factors (e.g. Eby et al., 2022;Gibb et al., 2020), the importance of immunogenetic diversity cannot be overstated. Population (immuno-)genetic diversity is an essential aspect of biological diversity, determines host competence to harbour and transmit pathogens and, hence, safeguards against future pandemics. Yet, in spite of the presence of CoVs in Ghanaian bats, locals regularly enter the bat caves (Anti et al., 2015;Leroy et al., 2009;Monadjem et al., 2007;Nkrumah et al., 2021). For policymakers and conservationists, this means that in places where stark dependencies between the local community and wildlife still exist, alternative nutritional, cultural and economic incentives need to be created to sustainably curb humanwildlife contact.

ACK N O WLE D G E M ENTS
We would like to thank Tobias Lenz and Mark Gillingham for feedback on the codon usage analysis. Special thanks to Alexandre Courtiol, who was imperative to implementing the R code for the codon usage analysis. We thank the Ghanaian chiefs and community leaders for their support during sampling. Open Access funding enabled and organized by Projekt DEAL.

FU N D I N G I N FO R M ATI O N
The research was funded by the German Research Foundation (DFG SO 428/17-1, with samples and metadata obtained by DR 772/3-1 and 7-1, KA1241/18-1, TH 1420/1-1).

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare no conflict of interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
MHC sequences and data is accessible on GitHub (https://doi. org/10.5061/dryad.h9w0v t4p4).

O PEN R E S E A RCH BA D G E S
This article has earned an Open Data Badge for making publicly available the digitally-shareable data necessary to reproduce the reported results. The datasets used in this study are available in Dryad Repository (https://doi.org/10.5061/dryad.h9w0v t4p4).

B EN EFIT-S H A R I N G S TATEM ENT
A research collaboration was developed with scientists from Ghana where samples originated. All collaborators were included as coauthors. Knowledge gathered from the research will be shared with both the scientific community and general public. Our group is committed to international scientific partnerships, as well as institutional capacity building. We also provide free access to data and code.