Shared and unique microbes between Small hive beetles (Aethina tumida) and their honey bee hosts

Abstract The small hive beetle (SHB) is an opportunistic parasite that feeds on bee larvae, honey, and pollen. While SHBs can also feed on fruit and other plant products, like its plant‐feeding relatives, SHBs prefer to feed on hive resources and only reproduce inside bee colonies. As parasites, SHBs are inevitably exposed to bee‐associated microbes, either directly from the bees or from the hive environment. These microbes have unknown impacts on beetles, nor is it known how extensively beetles transfer microbes among their bee hosts. To identify sets of beetle microbes and the transmission of microbes from bees to beetles, a metagenomic analysis was performed. We identified sets of herbivore‐associated bacteria, as well as typical bee symbiotic bacteria for pollen digestion, in SHB larvae and adults. Deformed wing virus was highly abundant in beetles, which colonize SHBs as suggested by a controlled feeding trial. Our data suggest SHBs are vectors for pathogen transmission among bees and between colonies. The dispersal of host pathogens by social parasites via floral resources and the hive environment increases the threats of these parasites to honey bees.


| INTRODUC TI ON
The small hive beetle (Aethina tumida Murray, 1867, hereafter SHB) is a honey bee nest parasite belonging to the family Nitidulidae (sap beetles; c. 4,500 species), whose members feed mainly on decaying vegetable matter, over-ripe fruit, or sap (Mckenna et al., 2015).
Unlike other plant-feeding beetles, SHBs can survive on fruit but thrive on resources found in honey bee colonies (Cuthbertson et al., 2013;Neumann & Elzen, 2004). SHB larvae are the most damaging stage for bee hives, by tunneling through combs and causing honey to ferment (Hood, 2004). These infestations can be destructive to wax combs, stored honey, and pollen. So far, the yeast Kodamaea ohmeri is known to be associated with SHBs, causing damage to the colony by fermenting stored nectar and serving as a biomarker to attract other SHBs (Benda, Boucias, Torto, & Teal, 2008). Additional symbiotic microbes associated with SHBs have not yet been described. In contrast, several symbiotic bacteria have been reported from the Asian longhorned beetle, including those that facilitate plant cell wall digestion (Scully et al., 2013), leading to insights into how these microbes impact digestion and beetle health.
Honey bee gut bacteria are dominated by nine species/clusters, some of which are likely to be involved in honey and pollen digestion, along with many low-frequency opportunistic microbes (Kwong & Moran, 2016;Powell, Martinson, Urban-Mead, & Moran, 2014;Raymann & Moran, 2018). As SHBs rely on food sources stored by their honey bee hosts, we predicted that SHBs might acquire honey bee-associated microbes, which could aid in food digestion. In addition, SHBs maintain their own sets of bacteria that could aid in digestion, improving development inside the colony and when they exit as late-stage larvae to finish development.
In this study, we conducted metagenomic sequence de novo assembly to identify microbes found in larval and adult life stages of SHBs. We then confirmed those microbes using a deep RNA-seq data set. We further conducted controlled feeding trials to determine whether candidate microbes can colonize SHBs. We have identified microbes that might facilitate the defense and development of the SHBs. We also found bee-associated bacteria and viruses residing in SHBs. These results shed light on beetle microbe communities and help identify risks to both bees and beetles from a communal existence, as well as complex pathogen transmission routes in this ecosystem.

| Beetle collection and DNA extraction
SHBs were collected from the states of Louisiana and Maryland, USA. DNA was extracted from three adult beetles for Illumina HiSeq paired-end sequencing in 2011. Additionally, DNA was extracted from 150 SHB larvae for PacBio sequencing in 2014. These two data sets are not related, and the sequencing was conducted at the University of Maryland. These non-sterile adult and larval small hive beetles were scrutinized to identify microbes shared with bees, microbes unique to SHB, and microbes picked up from the hive or external (soil) environment. These two sets of DNA sequencing reads were previously used to assemble the SHB genome (https ://www. ncbi.nlm.nih.gov/assem bly/GCF_00193 7115.1/). For detailed DNA extraction and sequencing protocol, see Evans et al. (2018. Due to extremely deep sequence coverage (over 500X SHB genome coverage), we were able to accurately explore the microbial community associated with SHBs. Pooled, equimolar RNA sequencing reads of eggs, larvae, and adult beetles were previously used to construct the SHB transcriptome (over 500x SHB transcriptome coverage, as described in Tarver et al., 2016). This RNA resource was used to assess the transcriptional activity of these microbes in SHB. Both DNA and RNA sequences were previously deposited at NCBI-Bioproject PRJNA256171.

| Metagenomic analysis of beetleassociated microbes
Ilumina reads were quality checked with Fastqc (http://www.bioin forma tics.babra ham.ac.uk/proje cts/fastq c/), and PacBio reads were error corrected with Illumina reads using proovread (Hackl, Hedrich, Schultz, & Forster, 2014). DNA and RNA reads were first aligned to the SHB genome using BWA (version 0.7.13) and Tophat2 (version 2.0.13), respectively (Kim et al., 2013;Li & Durbin, 2009). Reads aligned to the SHB genome were removed. After this filtering, 96 million Illumina DNA reads, 137 million Illumina RNA reads, and 247,186 PacBio reads (~870 million nucleotides) were maintained for microbial identification. Initially, the unmapped reads were used to screen microbial species with fully sequenced genome assemblies using Kraken with standard databases, which is designed to align short sequencing reads to sequenced microbe genomes (Wood & Salzberg, 2014) (Supporting Information S1). Kraken output files were viewed using Krona (Wood & Salzberg, 2014) (Appendix Figure A1, Figure A2 and Figure A3). In order to reduce numerous false-positive assignments of K-mers (subset of a read) from Kraken, a de novo metagenomic assembly was produced using unmapped Illumina DNA reads by metaSPAdes assembler (version 3.10.1) with default setting (Nurk, Meleshko, Korobeynikov, & Pevzner, 2017). The assembled contigs and unmapped PacBio long reads were used to query the Embl, Unigene, Est, Gss, Htc, Pat, RefSeq, Htg, and Tst databases using BLASTN. Best hits were tallied for searches with alignment significance of p < 0.001. Only microbes confirmed by both Kraken and the assembled contigs were kept. In order to identify bee-associated microbes found in SHBs, the unmapped DNA and RNA reads were aligned to the HoloBee database, a curated resource for microbes associated with honey bees (https :// data.nal.usda.gov/datas et/holob ee-datab ase-v20161), using BWA (version 0.7.13) and Tophat2 (version 2.0.13), respectively. Again, candidate matches were aligned against both assembled contigs and unmapped PacBio reads to reduce false-positive assignments ( Figure 1).
HoloBee-Barcode uses a variety of markers as appropriate for each taxonomic group (Supporting Information S2). Complete 16S ribosomal RNA was used for bacteria. Barcode markers for fungi are less F I G U R E 1 Using three independently sequenced SHB samples to identify the associated microbes, adult and larval beetle DNA reads were first aligned to all sequenced microbes genomes using Kraken and validated with RNA sequencing reads. Then, the adult and larvae DNA reads were aligned to HoloBee database and again validated with RNA sequencing reads. The adult DNA reads were de novo assembled, and the contigs were aligned to Embl, Unigene, Est, Gss, Htc, Pat, Refseq, Htg, and Tst databases to further validate the species/ gene origin of the contigs definitive, and ribosomal RNA internal transcribed spacer region (ITS), including ITS-1, 5.8S, and ITS-2, was used via Holobee database. The majority of barcodes for metazoan taxa are based on the mitochondrial locus Cytochrome C oxidase subunit I. Read counts were normalized with trimmed means of M-values (TMM) using edgeR (Robinson, McCarthy, & Smyth, 2010). Over all, there are two steps to reduce false-positive assignment of the identified microbes. First, the microbes identified from the Kraken database and Holobee database must be supported by both DNA and RNA reads. Second, the identified microbes must show significant hit when blasting the assembled de novo contigs to Embl, Unigene, Est, Gss, Htc, Pat, Refseq, Htg, and Tst databases (p < 0.001; Supporting Information S4).

| Verification of the identified microbes with qPCR
To further validate the accurate assignment of microbes from sequencing, a set of microbes (Choristoneura occidentalis granulovirus, Kodamaea ohmeri, Deformed wing virus, Gilliamella apicola, and Snodgrassella alvi) was selected for qPCR validation. To accomplish this, 12 adult beetles were freshly collected from apiaries near Baltimore, Maryland, in June 2018. DNA was extracted from individual beetles, and each of 3 beetles from an apiary was pooled for qPCR analysis. For detailed protocol and results, see Appendix and Supporting Information S4.

| Colonization of honey bee-associated microbes in SHBs
We further studied whether the selected set of microbes (Choristoneura occidentalis granulovirus, Kodamaea ohmeri, Deformed wing virus, Gilliamella apicola, and Snodgrassella alvi) can colonize small hive beetles. Accordingly, an additional 10 adult beetles were collected from the honey bee hives in Beltsville, Maryland, in September 2018. Those 10 beetles were feed with sugar water for 7 days, without introduction of any bee hive products. We hypothesize that if the microbes remained in place under this controlled diet, they can could truly colonize SHBs, instead of being merely transients collected from bee hive products. After 7 days feeding, each SHB was dissected into head thorax and abdomen sections. Then, the same body sections from five SHBs were pooled for RNA/DNA extraction, to determine specific tissue colonization of microbes. Detailed DNA extraction, RNA extraction, and qPCR protocols, along with the primers and results, are described in Supporting Information S3 and S4.

| Identification of microbes from the small hive beetle
In total, 66 and 23 different microbe species were found from SHB larvae (2 archaea, 55 bacteria, and 9 viruses) and adults (22 bacteria and 1 viruses), respectively (Appendix Table A1). Of those, 14 bacteria were shared between SHB larvae and adults, including 9 putatively beneficial bacteria ( Table 1). The bacteria Gluconobacter oxydans, Candidatus Pantoea carbekii, secondary endosymbiont of Heteropsylla cubana, and Lactococcus lactis were found in SHB larvae, as well as a toxin-secreting bacterium "Candidatus Profftella armatura".

| Bee-associated microbes found in the small hive beetle
As the SHB feeds on honey and pollen in honey bee colonies, these beetles are expected to receive microbes (pathogenic and symbiotic) from resident honey bees and hives. We used the Holobee database,

| Verification of the microbes with qPCR
Out of the five selected microbes, only Choristoneura occidentalis granulovirus was not confirmed, neither from adult nor larval SHBs (Table 3, Appendix

| Controlled diet analysis of SHB microbes
Deformed wing virus persisted in beetles fed under a controlled diet. Gilliamella apicola and Snodgrassella alvi were found in beetles collected from colonies but were absent after the controlled diet trials. The yeast K. ohmeri was highly abundant and constantly identified both before and after the controlled diet trials. Choristoneura occidentalis granulovirus was not found in beetles either before or after diet trials.

| SHB unique microbes
Candidatus Pantoea carbekii is a known mutualism of plant-feeding insects, which may facilitate survival and development by providing essential nutrients (Kenyon, Meulia, & Sabree, 2015). In our data, this bacterium was found in larval SHB samples, perhaps supporting the development of SHB by supplying nutrition. Protective bacteria were also found associated with SHBs. Candidatus Profftella armatura secretes polyketide toxins to protect plant-feeding insect hosts from predators (Nakabachi et al., 2013), and it is conceivable that SHBs benefit from this bacterium when facing predators inside and outside the nest. For the Asian longhorned beetle, ten genera of bacteria were linked with lignocellulose and hemicellulose degrada- quite different environmental condition compared to the bee hive.
The described microbes were supported by independent data sets, reducing the chance that those microbes are falsely assigned.

| Honey bee-associated microbes found in SHBs
Out of the nine dominant bacteria species/clusters found in honey bees (Moran, 2015), four were found in SHBs, including three proteobacteria Gilliamella apicola, Frischella perrara, and Snodgrassella alvi, and one Firmicutes bacteria Lactobacillus kunkeei. The bacterium G. apicola facilitates pollen digestion and has a syntrophic effect with S. alvi that is very abundant in our study (Kešnerová et al., 2017). Acquiring this core set of honey bee bacteria arguably could help the beetle degrade pollen cell walls and digest sugars found in stored honey (Kwong & Moran, 2016

ACK N OWLED G M ENTS
We thank James P. Tauber for providing K. ohmeri primer se-

CO N FLI C T O F I NTE R E S T S
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
JDE and QH designed the work, performed metagenomic analysis, and wrote the manuscript. DL performed qPCR validation and analyzed the data.

E TH I C S S TATEM ENT
None required.

DATA AVA I L A B I L I T Y S TAT E M E N T
DNA and RNA sequencing reads were previously deposited at NCBI-BioProject PRJNA256171.