Host and microbial factors influence bacterial colonization of the honey bee gut
Silvia Moriano-Gutierrez1,2,#, Aurélien Pirat1, Audam Chhun1, Aiswarya Prasad1, Alexandra Szigeti1, Méline Garcia1, Lucie Kesner1, Florent Mazel1, Philipp Engel1,#
1 Department of Fundamental Microbiology, University of Lausanne, Lausanne, Switzerland. 2 Institute of Agrochemistry and Food Technology, National Research Council, València, Spain.
Introduction
The animal gut microbiome fulfills key functions for the host in nutrition, development, defense and even behaviour1–7. While gut symbionts are horizontally transmitted, they usually exhibit a high degree of host specificity8–11. Understanding the mechanisms that underlie microbiota assembly and symbiont specificity remains a central question in microbiome research. Multiple factors, including host filtering (host-derived constraints on microbial establishment and/or abundance), microbe-microbe interactions (competition/facilitation), and dispersal limitation, are known to contribute8,9,12,13, but their relative influence, particularly across microbial strains with different evolutionary histories, is still not fully understood. Evidence from other host systems shows that colonization potential is not determined by isolation source alone: microbes from diverse environments can establish in the gut when they possess traits suited to that habitat14. This raises the question of whether similar trait-based mechanisms explain colonization by non-native bacteria, and how they interact with host-specific constraints and resident microbiota.
The assembly of animal gut microbiota is governed by a complex interplay of host and microbial factors that collectively shape which microbes establish and persist. From the host side, immune responses, gut physiology, and nutrient availability act as selective filters that can promote native symbionts while excluding non-native taxa8. However, it remains unclear whether native and non-native taxa differ enough for the honey bee’s innate immune system to differentially detect and constrain them, including those from closely related hosts. From the microbial perspective, colonization requires overcoming several ecological and physiological challenges: entering the host, reaching and replicating in specific gut niches, resisting immune defenses, and interacting with competing microbes15. Traits such as antimicrobial peptide (AMP) resistance, metabolic compatibility and immune evasion all contribute to whether a strain can establish in the gut16–22. Additionally, ecological interactions between microbes, including competition (e.g., nutrient exclusion, niche preemption) and facilitation (e.g., cross-feeding, immune modulation), can shape community outcomes19,23–26. These interactions are further constrained by behavioral and ecological barriers of hosts, such as dietary habits or restricted transmission routes, which limit dispersal and reinforce host-specific assemblages10,27,28.
Honey bees ( Apis mellifera ) serve as a powerful model organism for investigating factors underlying host-microbe specificity. As social corbiculate bees, they harbor a simple yet highly specialized gut microbiota that can be readily studied under controlled laboratory conditions29. This community comprises approximately 15-20 bacterial species across a few dominant genera30–32. Adult worker guts are consistently dominated by five “core” genera: Gilliamella , Snodgrassella , Lactobacillus , Bombilactobacillus and Bifidobacterium . Additional bee-associated genera, such as Frischella , Bartonella , and Commensalibacter , are frequently detected but are generally considered non-core, with prevalence varying among individuals and colonies33. Opportunistic colonizers, such as environmentally acquired commensals or pathogens (e.g. Serratia or Hafnia ) are also occasionally detected but occur relatively rarely34. This characteristic gut microbiota is primarily acquired through social transmission shortly after adult emergence, via contact with nestmates and hive material35,36. Although the gut microbiota of other social bees, such as other Apis species, bumble bees, and stingless bees, often include members of the same genera, these bacteria are typically represented by host-specific species or strains. Moreover, these symbionts are rarely found outside of the gut of bees, suggesting a high degree of host specificity33. However, the ecological processes and molecular mechanisms underlying the observed specificity are not fully understood. Experimental studies with gnotobiotic bees have demonstrated that bacteria of Snodgrassella , Gilliamella , and Lactobacillus generally exhibit greater colonization success in their native bee host than in other bee species30,32. Bacterial colonization of the bee gut activates host immune responses, including AMP production and modulation of key signaling pathways such as the Toll and Imd pathways37–39. In one specific case, it was shown that a non-native, but not a native strain, of the genus Gilliamella triggers the production of reactive oxygen species thereby partially hindering gut colonization39. To date, however, most studies have examined only a small number of strains, sometimes belonging to different species and often limited to a single “native vs. non-native” comparison. As a result, it remains unclear whether gut colonization is truly restricted to the recurrent core genera of social bees, or whether additional, underexplored lineages can also establish under certain conditions. Moreover, substantial variation in colonization efficiency, metabolic capacity, and immunogenicity is known even among closely related strains within core taxa40,41. Such intra- and interspecific diversity highlights the need for phylogenetically broad, strain-resolved experimental surveys that move beyond a few model isolates to broaden inference about the true breadth of colonization potential across ecologically defined bacterial groups.
In this study, we investigated the factors influencing bacterial colonization success in the honey bee gut using gnotobiotic bees. We assembled a diverse panel of 56 phylogenetically and ecologically distinct bacterial strains, grouped into four categories: (i) native symbionts belonging to the typical genera found in the honeybee gut microbiota, (ii) opportunistic colonizers of the honey bee gut, (iii) gut symbionts from other bee species, and (iv) non-bee environmental isolates. By evaluating colonization performance in both mono-association and community settings, and by characterizing AMP sensitivity, metabolic profiles, and interbacterial interactions, we identify bacterial traits and host factors that shape colonization outcomes. This integrated approach provides new insights into the complex and multifactorial nature of microbiota assembly and host specificity in animal symbioses.
Materials and methods
Generation of gnotobiotic bees
Late-stage pupae with pigmented eyes and light grey cuticle were removed from brood frames and incubated for 3 days at 35 °C and 75% humidity in sterile boxes, receiving sterile 1:1 sucrose solution. Upon emergence, bees were randomly assigned to one of four groups: microbiota-depleted (MD), colonized with a synthetic (or defined) community (SC) of native symbionts, monocolonized with individual strains (S), colonized with SC plus an additional strain (SC + S). Strains, detailed in Supplementary File 1, were grown, washed, standardized to OD₆₀₀ = 1.0, and stored in glycerol at -80 °C. The 14-member SC was assembled to represent the dominant worker gut symbiont genera of A. mellifera and was selected based on (i) assignment to the most prevalent genera present in the gut of A. mellifera , and (ii) reproducible colonization in gnotobiotic bees. The 14 strains comprising the SC are listed in Supplementary File 1 under category Native. For colonization, bees were fed 5 μL of bacterial suspension, of the glycerol stocks stored at -80 °C, in PBS:sucrose. S treatments received a single strain (OD₆₀₀ = 0.1); SC included 14 native strains mixed at an OD₆₀₀ = 0.1; SC + S combined SC with one added strain. MD bees received mock inoculum (1:1 PBS-sucrose solution). Guts were dissected seven days post-inoculation and stored at -80 °C. Seven days post-inoculation is commonly used in gnotobiotic honey bee assays because gut colonization and community structure stabilize within the first week of adulthood42. The experiment included three biological replicates using pupae from three different hives to account for across-colony variation.
DNA extraction from honey bee gut tissue
DNA was extracted from honey bee gut samples using bead-beating followed by magnetic bead-based purification. Samples were kept on ice and processed alongside the negative controls. Each sample was lysed with ATL buffer containing Proteinase K (2 mg/mL) and mechanically disrupted using a bead-beater (6 m/s, 45 s). After incubation at 56 °C for 1 h with agitation, lysates were cleared by centrifugation, and DNA was purified using CleanNGS magnetic beads. DNA binding, ethanol washes, and elution were performed in a 96-well plate format. Final DNA eluates (40 μL) were stored at -20 °C for downstream analysis.
RNA extraction from honey bee gut tissue
Total RNA was extracted from whole gut samples using the RNeasy PowerMicrobiome Kit (Qiagen), following the manufacturer’s instructions. Immediately after extraction, samples were treated with TURBO DNase (Thermo Fisher Scientific) to eliminate residual genomic DNA. RNA concentration for each sample was quantified using the Qubit RNA Broad Range (BR) Assay Kit (Invitrogen), according to the manufacturer’s protocol.
Bacterial quantification by quantitative PCR (qPCR)
qPCR was performed on a QuantStudio™ 5 system (Applied Biosystems) in 96- or 364-well plates, using 10 μL reactions with 0.2 μM primers, 1× SYBR® Select Master Mix, and 1 μL DNA template. Thermal cycling included 50 °C for 2 min, 95 °C for 2 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. Each sample was run in triplicate. Total bacterial load, host gene expression, and strain-specific abundance were assessed using universal, species-specific, and host reference primers (see Supplementary File 2). Universal 16S rRNA primers were used only for MD and SC controls to assess background signal and total bacterial load, whereas strain-specific primers were used in inoculated treatments to quantify inoculated strain abundance and prevalence. Primer design followed Primer3plus; quantification methods are described in Kešnerová et al43.
Host immune expression by qRT-PCR
For host immune gene expression, cDNA was synthesized from gut RNA using SMART MMLV reverse transcriptase with both Oligo(dT) and random hexamers. qRT-PCR followed MIQE guidelines, including no-RT and no-template controls. Standard curves ensured amplification efficiency, and reactions used 12.5 ng cDNA in triplicate. Melting curve analysis confirmed specificity. Gene expression was normalized to the Apis mellifera RPS18 reference gene. Details on all primer sequences and target genes are provided in Supplementary File 2.
DNA isolation for genome sequencing
Genomic DNA for Illumina sequencing was extracted using the FastPure Bacterial DNA Isolation Mini Kit (Vazyme), following the manufacturer’s protocol for Gram-positive bacteria. Illumina sequencing libraries were prepared using the Nextera DNA Flex Library Prep Kit (Illumina), following the guidelines in the Illumina Reference Guide. Library indexing, dilution, and denaturation were performed according to the Illumina Index Adapters Pooling Guide and the MiniSeq System Denature and Dilute Libraries Guide. Final libraries were quantified using a double-stranded DNA fluorescent dye assay and sequenced on a MiniSeq high-output flow cell (150 bp paired-end reads).
Genome assembly and functional annotation
Raw Illumina reads were quality-checked using FastQC v0.11.944 and trimmed with Trimmomatic v0.3945 using the following parameters: PE -phred33 AllIllumina-PEadapters.fa:3:25:7 LEADING:9 TRAILING:9 SLIDINGWINDOW:4:15 MINLEN:60. Trimmed reads were assembled de novo using SPAdes v3.15.2 with the --careful option46. Genome completeness and contamination were evaluated using CheckM v1.0.1347 (Supplementary File S7). Genomes were annotated with Prokka v1.1348. Open reading frames (ORFs) were identified on all assembled scaffolds of each sample using Prodigal49. ORFs marked partial or shorter than 300 bp were removed. The remaining ORFs were then annotated using DRAM50 to obtain KEGG ortholog (KO) identities.
Phylogenetic analysis
A phylogenetic tree of the 56 bacterial species and strains used in colonization experiments was constructed using OrthoFinder v2.5.4. Orthologous gene groups were identified across all genomes, and species tree inference was performed using the STAG (Species Tree from All Genes) algorithm51, as implemented in OrthoFinder52. This approach integrates gene tree topologies to produce a robust consensus species tree.
KEGG module annotation and metabolic profiling
Genomes were annotated and KEGG module completeness quantified per strain, generating a strain × module matrix (Supplementary File 7). Modules with ≥80% completeness were considered complete. For each strain, we calculated module richness and module-level completeness values for association testing. PCA was performed on the centered and scaled matrix. Module–abundance associations were tested with per-module mixed-effects models (log_median ~ module + Category + Colonization_mode + (1 | genus)) and FDR correction. Module richness was tested against abundance with lme4::lmer(log_median_rank ~ Number.of.modules + Category + (1 | genus)); when genus-level variance was ~0, the corresponding fixed-effects model is also reported. Differences in module richness among categories were tested with number of modules ~ Category + (1 | genus).
Antimicrobial peptide resistance assays
To evaluate bacterial sensitivity to AMPs, resistance assays were conducted using a 12-point, two-fold serial dilution series of each AMP, ranging from 100 μg/mL to 0.0487 μg/mL. AMPs were synthesized by Shanghai Royobiotech Co., Ltd., with peptide sequences detailed in Supplementary File 6. A subset of strains was selected for AMP screening to represent each ecological category and to span the range of in vivo colonization outcomes; strains were additionally prioritized based on their ability to grow reproducibly and reach sufficient biomass under the in vitro assay conditions. Each bacterial strain was tested in triplicate across all peptide concentrations. Bacterial cultures were inoculated into 96-well microtiter plates containing AMP dilutions in 50% growth medium and incubated under standard conditions (Supplementary File 1). Bacterial growth was monitored by measuring optical density at 600 nm (OD₆₀₀) over time to assess biomass accumulation and peptide-induced growth inhibition. Growth inhibition was evaluated relative to untreated controls. Normalized area under the curve values were compared to a control mean of 1 using one-sample t-tests at each concentration and condition. P-values were adjusted for multiple testing using the FDR method. Growth was considered significantly inhibited at a given concentration if q < 0.05. A sensitivity score from 0 to 12 was assigned to each strain-peptide combination, representing the number of concentrations at which significant growth inhibition was detected. A score of 12 indicated growth inhibition even at the lowest concentration tested, while a score of 0 indicated no significant inhibition at any concentration.
Statistical analyses
All statistical analyses were performed in R (version 4.2.2). Data are presented as mean ± standard deviation for approximately normally distributed data and as median with 95% confidence intervals otherwise. For parametric models, assumptions were checked on the model residuals. Normality of residuals was assessed using the Shapiro-Wilk test and visual inspection of Q-Q plots, and homoscedasticity was evaluated from residual-versus-fitted plots. Depending on the data distribution, statistical comparisons were performed using one-way ANOVA or Kruskal-Wallis tests. Proportion data were analyzed using chi-square tests for equality of proportions, with FDR correction for multiple comparisons where applicable. A p-value < 0.05 was considered statistically significant. Where applicable, p-values were adjusted for multiple comparisons. Sample size (n) refers to the number of independent biological replicates. Independent experimental replicates are indicated when applicable. Detailed information on statistical tests used is provided in the corresponding figure legends.
Results
Bacterial strains vary in their ability to colonize the honey bee gut regardless of their isolation source
Using gnotobiotic bees, we screened 56 taxonomically and ecologically diverse bacterial strains for their ability to colonize the midgut and hindgut of the Western honey bee ( Apis mellifera ), either alone or in the presence of a synthetic community representing eight predominant gut genera. The strains belonged to 25 genera spanning Gamma- and Betaproteobacteria, Actinobacteria, and Firmicutes (Figure 1A) and fell into four ecological categories (Figure 1B, Supplementary File 1): (i) Native, isolated from A. mellifera and representing predominant gut genera; (ii) Opportunistic colonizers, isolated from A. mellifera but not belonging to predominant genera and typically detected at low abundance or prevalence; (iii) Non-native bee-associated, belonging to the same prevalent genera as native strains but isolated from other bee species and representing distinct bacterial species; and (iv) Other microbiomes, originating from non-bee environments such as soil, fish, and human or mouse guts. Seven days after inoculation, loads of all 56 strains were quantified by strain-specific qPCR (Figure 1A-C, Figure S1). Universal 16S rRNA qPCR of MD and SC controls (Figure S1) showed low background signal in MD bees and significantly higher bacterial loads in SC bees. Both the bacterial load per gut and the proportion of bees with levels exceeding the initial inoculum varied substantially across strains, ranging from 103 to 1010 cells per bee (Figure 1C, Supplementary File 2) and from 0% to 100%, respectively (Figure S2). On average, the highest colonization loads were observed for strains derived from the native gut microbiota of Apis mellifera , followed by non-native bee-associated strains and opportunistic colonizers. In contrast, isolates originating from non-bee environments exhibited the lowest colonization levels (Figure 2A). Similarly, the proportion of colonization was significantly higher for native strains compared to those derived from other bee species (≈21% lower relative to native) or from other microbiomes (≈67% lower relative to native) (Figure 2B). Opportunistic colonizers of Apis mellifera reached a similar proportion of colonization success as native strains, though with a lower bacterial load. When introduced alongside a synthetic community representing the native microbiota, the differences in colonization patterns across the four ecological groups were comparable to those observed in the monocolonizations (Figure S2); however, both bacterial load and proportion of colonization were generally reduced, likely due to competitive interactions with the native community (Figure 1C, Figure 2C). Only one native strain, Gilliamella ESL0177, exhibited enhanced growth in the presence of the synthetic community compared to monocolonization. While this strain colonized poorly when introduced alone (Figure 1C, Figure S2), its bacterial load increased substantially in a community context, suggesting positive interactions with other native microbiota members facilitating its colonization.
Given the observed variation in colonization among strains, we tested whether phylogenetic relatedness to native gut symbionts of Apis mellifera could account for differences in colonization. Consistent with this hypothesis, we found a significant negative correlation between phylogenetic distance to the closest native strain and colonization abundance among non-native strains for both monocolonization and community context (Figure 2D, Figure S3). Overall, these findings demonstrate that while the ability to colonize the honey bee gut is not limited to native strains, native symbionts generally exhibit higher colonization loads.
KEGG-based metabolic profiles reveal category-specific rather than universal colonization signals
To investigate whether metabolic capacity underlies differences in colonization between strains, we evaluated the completeness of KEGG functional modules across all bacterial strains. A total of 40 KEGG metabolic pathways were identified, and module completeness was quantified per strain (Figure S4, supplementary File S4). A PCA of KEGG module completeness did not separate strains by colonization performance: strains with high and low gut loads were intermixed (Figure 3A), and the same pattern was observed for colonization efficiency (% colonization; Figure S5). Hierarchical clustering based on KEGG modules grouped several Native and Other bee species strains together, especially those with well-represented central carbohydrate metabolism, but this clustering did not reflect colonization outcome (Figure S6). We then asked whether specific metabolic functions were linked to colonization performance. A KEGG module-specific mixed-effects scan for colonization abundance and efficiency identified only a small number of modules with significant associations (p < 0.05), and these effects were modest. Positively associated modules were mostly amino-acid-related, with lysine biosynthesis linked to higher abundance and tyrosine, theorine, and leucine pathways linked to higher colonization efficiency (Figure 3B, supplementary file S4). In contrast, several stress- or envelope-related modules showed negative coefficient values in the abundance model. Modules linked to cell-wall remodeling or antibiotic-resistance-type functions (e.g. vancomycin-type D-Ala-D-Lac remodeling) and to siderophore/iron acquisition were associated with modest but significant reductions in bacterial load. This pattern was not observed for colonization efficiency, where the main signals were amino acid and other anabolic pathways with positive effects (Figure 3B, supplementary file S4). Although metabolic completeness did not universally predict colonization capacity across all ecological groups, we observed significant category-specific trends. Specifically, a positive correlation was found between the number of high-completeness KEGG modules and median colonization abundance for strains from Other microbiomes , while a negative correlation was observed for Native strains (Figure 3C). However, the overall number of KEGG modules with ≥80% completeness did not significantly differ between ecological categories (Figure 3D).
Differential activation of host immune effectors by individual gut bacterial strains
To evaluate how different bacterial colonizers influence host immune responses, we mono-associated MD honey bees with 25 of the 56 strains and measured the expression of seven immune-related genes ( Apidaecin1 , Abaecin , Defensin1 , Defensin2 , Hymenoptaecin , IRP30 , Lysozyme ) in the gut using qPCR seven days post-colonization. Strains were selected based on their ability to reach colonization levels >106 bacteria per gut (Figure 1C) and to represent each of the four distinct ecological categories. Expression levels were evaluated relative to MD controls. Overall, we observed relatively few significant changes in immune gene expression between MD bees and those colonized with bacteria. For instance, only Apidaecin 1 and Defensin 2 were consistently upregulated in bees colonized with the full native community compared to MD controls. Likewise, only a limited number of monocolonizations led to detectable changes in gene expression (Figure S7A, Supplementary File S5). AMP and Lysozyme expression levels varied considerably across individual bacterial strains, but these differences did not segregate by ecological category; strains from the same group (e.g. native Gilliamella , native Lactobacillus ) often elicited divergent responses (Figure S7A). While certain non-native strains triggered elevated expression of Apidaecin 1 , Defensin 2 , and Lysozyme , others within the same group did not. Similarly, some native strains also elicited measurable immune responses, while others did not. No consistent expression patterns were observed for Hymenoptaecin or IRP30 across any colonizer group. Exceptions to this general trend included Defensin 2 , which was more frequently upregulated by native strains, and Lysozyme , which showed the opposite trend, being more often induced by non-native strains (Figure S7B). However, overall, immune gene expression responses appeared primarily strain-specific rather than determined by ecological origin.
To further examine whether immune gene expression levels are associated with colonization efficiency, we correlated the median abundance of each bacterial strain in monocolonization with expression of the seven immune genes. No significant correlations were observed (Figure S8), though Lysozyme showed a weak negative trend (R=-0.34, p=0.08). Thus, variation in host immune gene expression at seven days post-colonization does not reliably predict bacterial colonization efficiency.
Sensitivity of native and non-native strains to host AMPs.
To better understand the factors influencing colonization in the honey bee gut, we investigated bacterial sensitivity to host-derived AMPs, including Apidaecin1, Abaecin, Defensin1, Defensin2, and Hymenoptaecin, as these immune effectors are known to act selectively on microbes and could differentially inhibit strains depending on their resistance capabilities. To evaluate the potential role of host AMPs in shaping colonization outcomes, a subset of native and non-native strains was screened in vitro, and sensitivity scores were derived based on growth inhibition across AMP dilution series, with higher scores reflecting greater susceptibility (Figure 4A, Supplementary File 6). Gilliamella strains were among the most AMP-sensitive, with all five tested strains showing susceptibility to at least three AMPs. Similarly, all three Snodgrassella strains were sensitive to three or more AMPs, though their overall sensitivity scores were lower than those of Gilliamella . Among opportunistic colonizers, Hafnia also exhibited high sensitivity, being inhibited by at least three AMPs. Notably, strains of Apibacter and Bifidobacterium isolated from Apis cerana also showed broad AMP susceptibility, each susceptible to at least three antimicrobial peptides. Strains from both the Native and Other bee species categories exhibited higher AMP sensitivity scores compared to other groups, with strains from the Other bee species category showing the highest overall sensitivity (Figure 4B). We next examined whether AMP sensitivity influenced colonization efficiency and colonization success. The median abundance of AMP-sensitive strains was negatively correlated with total sensitivity score in both monocolonization and community contexts (Figure 4C). However, the percentage of colonization success events did not significantly correlate with AMP sensitivity (Figure 4C). Because immune tolerance is likely shaped by host adaptation, we investigated whether bacterial sensitivity to AMPs is related to their phylogenetic relatedness to native strains. We found that total AMP sensitivity scores were significantly negatively correlated with phylogenetic distance to native strains (Figure 4D), indicating that non-native bacteria more closely related to native symbionts tend to be more sensitive to host immune effectors.
Discussion
In this study, we examined how host filtering mechanisms contribute to the specificity of the gut microbiota in social bees. We tested an ecological and phylogenetically diverse set of bacterial strains, and found that, on average, native strains achieved higher bacterial loads and colonized a greater proportion of hosts in both mono-association and in the presence of the native microbiota. This is consistent with some previous work on host specificity, which report pronounced home-site advantages among bee gut core symbionts (e.g., Snodgrassella 30, Gilliamella 39, Lactobacillus 32). In mouse models, native Bacteroidetes and Firmicutes strains similarly show home-site advantages, outcompeting non-native counterparts even in hosts lacking an adaptive immune system10. Together, these findings suggest that intrinsic host factors preferentially support resident lineages. Despite this overall trend, we also observed substantial variation in colonization success even among closely related native strains, highlighting the importance of testing host specificity across a diverse set of strains.
Moreover, some bacteria from environments outside the animal gut were able to colonize honey bees to a certain extent, and even in the presence of the native microbiota, most of the non-native strains still achieved some level of colonization. However, in most cases, their colonization success was significantly reduced. This suggests that host filtering is not entirely strict, and also that non-native strains are not completely outcompeted by native strains when introduced at sufficiently high numbers. Related findings in other host systems show that environmentally derived or heterologous host-associated bacteria can colonize, and in some cases even dominate, gnotobiotic or germ-free hosts14,53.
Host-specific colonization of bee gut symbionts likely reflects interactions among ecological, bacterial, and host factors. In honey bees, early social acquisition favors nestmate-associated bacteria, and fecal transplants can transfer a gut-like microbiota to microbiota-depleted bees35. Consistently, reduced colonization of both native and non-native strains in the presence of the native community (Figure 2C) indicates strong competition for gut establishment. Resource limitation, niche preemption, and antagonism may therefore allow resident, socially transmitted strains to exclude environmental or non-native bacteria, reinforcing host specificity, as seen in other systems19,54,55.
Because the negative relationship between colonization load and distance to the closest native strain was also observed in monocolonization (Fig. 2D), it cannot be explained solely by competition with resident close relatives. Instead, phylogenetic proximity likely reflects host compatibility, that is, conserved traits enabling persistence and growth under host-derived constraints, consistent with the association between AMP sensitivity and colonization load (Fig. 4C-D). Together with the higher colonization success of native strains, this indicates that bacterial traits and host filtering are key determinants of host specificity. One possible factor is metabolic adaptation to diet- or host-derived nutrients in the honey bee gut. Although utilization of pollen-derived carbohydrates is an important trait of bee gut symbionts56,57, non-native strains with similar functional profiles did not necessarily colonize better (Figure 3, Figures S5-S6). Indeed, only a small subset of amino acid-related modules was positively, and only modestly, associated with colonization. In addition, the negative association between KEGG module richness and colonization observed among native strains (Fig. 3C) may partly be driven by the representation of specific bacterial lineages within this group, particularly high-colonizing Lactobacillus / Bombilactobacillus strains that show comparatively lower KEGG module richness (Fig. S5), consistent with genome streamlining reported for these host-specialized bee gut symbionts32,58. An important limitation of this analysis is that KEGG annotation provides only a coarse-grained functional profile of a bacterium, with many functions either omitted or described only at a general level. Module completeness may be underestimated in fragmented assemblies. Differences in gene regulation and in the expression of metabolic pathways may further modulate colonization success, irrespective of the mere presence of these functions within the genome.
Another important factor that may contribute to host filtering is the host’s immune system. In our study, host immune gene expression in response to gut colonization appeared to be primarily strain-specific rather than clearly associated with a particular ecological category of the tested strains (Figure S7A). Even among native taxa, closely related strains differed in their capacity to stimulate bee immunity, mirroring previous reports that some native Lactobacillus strains can markedly upregulate AMP genes, whereas others do not59. Apidaecin was often strongly induced by colonization with the full native community or even a single native strain, while other AMPs ( Abaecin , Defensin ) changed little38. A niche-matched contrast underscores this strain specificity: although Frischella perrara and Snodgrassella alvi both colonize the ileum, F. perrara triggers substantially stronger AMP and immune-pathway response than S. alvi 60. Such differences may reflect not only bacterial traits but also the specific gut region sampled, such as midgut, pylorus, or the full hindgut, each of which exhibits distinct patterns of bacterial colonization and immune activity36,61. Overall, we observed relatively few differences in immune gene expression between colonized and MD bees seven days post-inoculation, except for Abaecin and Defensin 2 (Figure S7A). This is consistent with findings from Kwong et al38., who similarly reported limited immune gene induction five days post-colonization. It is possible that by these relatively late time points following bacterial colonization, the immune response to bacterial exposure had largely subsided. Indeed, several previous studies which assessed earlier stages of colonization, such as 12 h or 24 h post-inoculation, reported the expression of additional immune-related genes through both the immune deficiency (IMD) and Toll signaling pathways37,39,59 that may be involved in the recognition of native versus non-native bacterial strains. The fact that we did not observe any significant correlation between immune gene expression and bacterial abundance (Figure S8) is consistent with the idea that immune responses to bacterial colonization are transient. This suggests that host immune tolerance to native symbionts may be established even at early colonization times and then maintained, such that differences in immune response between strains become less apparent at seven days post-colonization.
Another important factor that may contribute to host filtering is the host’s immune system. In our study, host immune gene expression in response to gut colonization appeared to be primarily strain-specific rather than clearly associated with a particular ecological category of the tested strains (Figure S7A). Even among native taxa, closely related strains differed in their capacity to stimulate bee immunity, mirroring previous reports that some native Lactobacillus strains can markedly upregulate AMP genes, whereas others do not59. Apidaecin was often strongly induced by colonization with the full native community or even a single native strain, while other AMPs ( Abaecin , Defensin ) changed little38. A niche-matched contrast underscores this strain specificity: although Frischella perrara and Snodgrassella alvi both colonize the ileum, F. perrara triggers substantially stronger AMP and immune-pathway response than S. alvi 60. Such differences may reflect not only bacterial traits but also the specific gut region sampled, such as midgut, pylorus, or the full hindgut, each of which exhibits distinct patterns of bacterial colonization and immune activity36,61. Overall, we observed relatively few differences in immune gene expression between colonized and MD bees seven days post-inoculation, except for Abaecin and Defensin 2 (Figure S7A). This is consistent with findings from Kwong et al38., who similarly reported limited immune gene induction five days post-colonization. It is possible that by these relatively late time points following bacterial colonization, the immune response to bacterial exposure had largely subsided. Indeed, several previous studies which assessed earlier stages of colonization, such as 12 h or 24 h post-inoculation, reported the expression of additional immune-related genes through both the immune deficiency (IMD) and Toll signaling pathways37,39,59 that may be involved in the recognition of native versus non-native bacterial strains. The fact that we did not observe any significant correlation between immune gene expression and bacterial abundance (Figure S8) is consistent with the idea that immune responses to bacterial colonization are transient.
regulating symbiont abundance and maintaining microbial homeostasis. Interestingly, Gilliamella strains were among the most sensitive strains, followed by those of the genus Snodgrassella. While this might seem paradoxical given their status as successful colonizers, it aligns with previous reports showing that Gram-negative symbionts, particularly Gilliamella, are more vulnerable to apidaecins38,69. This apparent sensitivity may not reflect a disadvantage, but rather a finely balanced immune-microbe interaction where AMPs act as regulators that maintain microbial homeostasis rather than eliminators of symbionts70–72, complementing ROS-mediated mechanisms that exclude maladapted strains11,39. Indeed, while AMP sensitivity correlated with bacterial abundance, it did not significantly predict colonization success or failure (Figure 4C, lower panel). Thus, AMPs seem to act in a graded, context-dependent manner that controls symbiont load rather than strictly permitting or blocking colonization, a mode of action also reported in other insects where AMPs confine resident symbionts to specific tissues (e.g. coleoptericin-A in Sitophilus73).
Finally, we observed a significant inverse relationship between AMP sensitivity and phylogenetic distance from native bee strains (Figure 4D). Strains more closely related to core symbionts of A. mellifera were more sensitive to host AMPs. Rather than selecting for broad resistance, the immune system may have evolved to support stable interactions with beneficial microbes by modulating their growth. Immune effectors can mediate not just defense, but the spatial and population-level dynamics of beneficial symbionts, a principle that appears conserved across diverse host taxa74.
It is important to recognize that AMP resistance is not the sole determinant of microbial fitness within the host gut. Some native strains that display sensitivity to honey bee AMPs may still thrive due to alternative or compensatory mechanisms. For instance, co-colonization or cross-feeding interactions may buffer AMP-sensitive strains against immune pressure, enabling persistence in the gut despite their inherent susceptibility. AMP-sensitive strains may persist when shielded or supported by compatible community members. This phenomenon has parallels in other systems. In D. melanogaster, AMP-sensitive strains are often outcompeted unless co-colonized with more immune-tolerant bacteria that can modulate host immunity or provide spatial refuge75,76. Similarly, in mammals, microbial persistence under immune pressure is often mediated by cooperative traits such as biofilm formation, cross-feeding, or immune evasion mechanisms77,78. Moreover, the host immune landscape is itself complex and dynamic. Natural polymorphisms in AMP genes among honey bee populations could lead to variability in the selective pressures experienced by microbes in different individuals or colonies.
Understanding these layers of complexity will be important for unraveling the full spectrum of host-microbe compatibility and microbial persistence mechanisms in the gut ecosystem. While we focused here on single peptide effects, natural host environments feature mixtures of AMPs that may act synergistically or antagonistically, further shaping community dynamics in complex and non-linear ways79. Moreover, other arms of the host immune system such as the DUOX-reactive oxygen species pathway, are activated by gut bacteria in insects and can limit microbial proliferation39,80,81. In addition, non-immune system factors, such as access to specific gut niches (pylorus vs ileum vs rectum), attachment sites and competition within these microenvironments, motility, chemotaxis, are likely to play to play a relevant role in host filtering, which we have not explored in this study.
Taken together, our findings support a model in which colonization success results from multifaceted processes with interplay between host immune filtering and microbial ecological dynamics. Early in adult life, socially transmitted core symbionts gain priority access to gut niches and establish a resident community that acts as an ecological filter, suppressing most incoming strains through competitive exclusion, with only rare facilitation. At the same time, host antimicrobial peptides impose a second selective pressure that limits bacterial proliferation in a taxon- and strain-dependent manner, shaping persistence and abundance rather than uniformly blocking establishment. Together, priority effects, microbial interactions, and host immune constraints help explain how honey bees maintain a stable, host-specific adult gut microbiota despite continual exposure to environmental and non-native bacteria.
Acknowledgments
We thank the Engel lab for valuable discussions and support throughout this work. In particular, we are grateful to Garance Sarton-Lohéac for generously sharing her genome assembly pipeline. We thank Vincent Sommerville for assistance with 3D-printing the bee cages. We also thank Konstantin Schmidt, Pascale Vonaesch, Christoph Keel, Yolanda Schaerli and Carlos Gustavo Nunes da Silva for providing bacterial strains used in this study. We thank the FBM décanat for the support provided during the parental leave of S.M.G. This work was funded by an SNSF Consolidator grant (‘GLOBEE,’ grant number 213860), and the NCCR Microbiomes, a National Centre of Competence in Research (grant number 180575 and 225148), all funded by the Swiss National Science Foundation.
Conflicts of interest
The authors declare no competing interests.
Data availability
All sequencing data and assembled genomes have been uploaded in the NCBI Sequence read archive repository under the accession number PRJNA1364849. All analysis code is available on GitHub: https://github.com/mogusi8/Host-and-microbial-factors-influence-gut-bacteria-in-honey-bees/tree/main . All other relevant data generated during this study are included in this article and its supplementary files.
Figures
Supplementary material
Figure S1: Total bacterial load in control groups (MD and SC) seven days post-inoculation. Total 16S rRNA gene copies per gut were quantified by qPCR using universal bacterial primers, in microbiota depleted (MD) bees and native synthetic community (SC) colonized bees. Mann–Whitney test, n=48. Significant label: p<=0.0001:****.
Figure S2. Bacterial colonization success of the honey bee gut, seven days post-inoculation, in monocolonization or in the presence of the native synthetic community (SC). A Phylogenetic tree of the 56 bacterial species and strains tested for colonization of the honey bee gut. The tree was inferred with OrthoFinder using the STAG algorithm51. B. Difference in abundance (expressed in log10 fold change) of each strain between monocolonization and in the presence of the native community. Mono and Comm data were obtained from independent bees; the fold-change shown connects the same strain measured under the two conditions. Wilcoxon rank-sum test. The p-values were adjusted using FDR method. Significance labels shown as p<=0.05: , p=<=0.01: , p<=0.001 , ns: not significant. C. Percent of colonization success for each strain in monocolonization (Mono) and in the presence of the native community (Comm) based on qPCR absolute abundance calculated as genome equivalents. The gut of a bee was considered colonized when the abundance was higher than in the initial inoculum (see material and methods). All colonization experiments were conducted with n=12 bees, obtained from three independent experimental replicates conducted with bees from different bee hives.
Figure S3: Fold change (FC) in the load of non-native strains in monocolonization relative to colonization in community, plotted against the phylogenetic distance to their nearest native strain. Phylogenetic distances are the horizontal distances in the phylogenetic tree (Figure 1A). R depicts Spearman correlation coefficient. Point colors indicate ecological category (as defined in Fig. 1A).
Figure S4: The presence and completeness of KEGG modules in the genomes of individual strains. Heatmap of percent of completeness for each strain. Supplementary file S4.
Figure S5: PCA of KEGG module completeness across strains. Principal component analysis was performed on KEGG module completeness profiles. Each point is a strain, coloured by its colonization efficiency in monocolonization. The shape indicates the isolation category.
Figure S6: A. The presence and completeness of KEGG modules in the genomes of individual strains categorized by isolation source. Heatmap of % completeness for each strain in each category. Grey: Native ; orange: Other bee species ; blue: Opportunistic colonizers ; green: Other microbiomes . B. Heatmap of the number of modules that are at least 80% complete for each pathway group. Top label shows bacterial loads categorized in for groups based on the abundance distribution in monocolonization (See materials and methods). C. Distribution of bacterial loads in honey bee guts during monocolonization. Strains are categorized based on colonization levels, defined by load percentiles: No colonization (below the 25th percentile), Low colonization (25th-50th percentile), Medium colonization (50th-75th percentile), and High colonization (above the 75th percentile).
Figure S7: Host antimicrobial peptide responses to colonization by individual bacterial strains. A Fold-change in expression relative to MD of six antimicrobial peptide (AMP) genes, Apidaecin1 , Abaecin , Defensin1 , Defensin2 , Hymenoptaecin , IRP30 , and Lysozyme , in the honey bee gut during monocolonization with the indicated bacterial strains. AMP expression was measured relative to microbiota-depleted (MD) bees. Bees colonized with the synthetic native community (SC) are shown as reference controls. Strains are grouped by category and host origin, with separate panels for native and non-native strains. Data are shown as individual points with boxplots indicating the median and interquartile range. Asterisks indicate strains that induced a significant change in expression relative to the MD control (p < 0.05, Wilcoxon rank-sum test, FDR-corrected, n=6). B, Summary of Defensin2 and Lysozyme expression levels across different strain categories in control (MD and SC) or monocolonization conditions. Asterisks indicate strains that induced a significant change in expression relative to the MD control (p < 0.05, Wilcoxon rank-sum test, FDR-corrected). No significant differences in expression were observed among categories for Apidaecin1 , Abaecin , Defensin1 , Hymenoptaecin , or IRP30 . C, Genus corresponding to each Strain ID for figure S8.
Figure S8: Antimicrobial peptide expression at seven days post-colonization does not predict bacterial colonization efficiency. Correlation between antimicrobial peptide (AMP) expression and bacterial colonization abundance in the honey bee gut. Scatter plots show the relationship between the median relative expression of seven AMP genes ( Abaecin , Apidaecin1 , Defensin1 , Defensin2 , Hymenoptaecin , IRP30 , and Lysozyme ) and the median abundance of bacterial strains across samples. Each point represents a bacterial strain. Blue lines indicate linear regression fits. Spearman correlation coefficients ® and associated p-values are shown for each gene. Regression lines are shown for visualization only; statistical support for associations is assessed using the Spearman correlation coefficients and p-values reported in each panel.
Figure S9. Area under the curve (AUC) of bacterial growth, normalized to the untreated control and presented on a logarithmic scale. The right panel shows the AUC values at a representative concentration of Abaecin. Each condition was tested in triplicate (n = 3). Statistical significance was assessed using a one-sample Wilcoxon test against the null hypothesis of no growth inhibition (normalized value = 1), with p-values adjusted for multiple comparisons using the false discovery rate (FDR) correction. Significance levels are indicated as follows: ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05.
Figure S10. Area under the curve (AUC) of bacterial growth, normalized to the untreated control and presented on a logarithmic scale. The right panel shows the AUC values at a representative concentration of Apidaecin 1a. Each condition was tested in triplicate (n = 3). Statistical significance was assessed using a one-sample Wilcoxon test against the null hypothesis of no growth inhibition (normalized value = 1), with p-values adjusted for multiple comparisons using the false discovery rate (FDR) correction. Significance levels are indicated as follows: ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05.
Figure S11. Area under the curve (AUC) of bacterial growth, normalized to the untreated control and presented on a logarithmic scale. The right panel shows the AUC values at a representative concentration of Apidaecin 1b. Each condition was tested in triplicate (n = 3). Statistical significance was assessed using a one-sample Wilcoxon test against the null hypothesis of no growth inhibition (normalized value = 1), with p-values adjusted for multiple comparisons using the false discovery rate (FDR) correction. Significance levels are indicated as follows: ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05.
Figure S12. Area under the curve (AUC) of bacterial growth, normalized to the untreated control and presented on a logarithmic scale. The right panel shows the AUC values at a representative concentration of Defensin 1. Each condition was tested in triplicate (n = 3). Statistical significance was assessed using a one-sample Wilcoxon test against the null hypothesis of no growth inhibition (normalized value = 1), with p-values adjusted for multiple comparisons using the false discovery rate (FDR) correction. Significance levels are indicated as follows: ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05.
Figure S13. Area under the curve (AUC) of bacterial growth, normalized to the untreated control and presented on a logarithmic scale. The right panel shows the AUC values at a representative concentration of Defensin 2. Each condition was tested in triplicate (n = 3). Statistical significance was assessed using a one-sample Wilcoxon test against the null hypothesis of no growth inhibition (normalized value = 1), with p-values adjusted for multiple comparisons using the false discovery rate (FDR) correction. Significance levels are indicated as follows: ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05.
Figure S14. Area under the curve (AUC) of bacterial growth, normalized to the untreated control and presented on a logarithmic scale. The right panel shows the AUC values at a representative concentration of Hymenoptaecin. Each condition was tested in triplicate (n = 3). Statistical significance was assessed using a one-sample Wilcoxon test against the null hypothesis of no growth inhibition (normalized value = 1), with p-values adjusted for multiple comparisons using the false discovery rate (FDR) correction. Significance levels are indicated as follows: ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05.
Supplementary File 1: Strain list.
Supplementary File 2: Primer list.
Supplementary File 3: qPCR abundances data.
Supplementary File 4: KEGG modules. Sheet 1: All KEEG modules detected. Sheet 2: Number of modules per strain with a >80% completeness, Sheet 3: Mix-model significant results with abundance data. Sheet 4: Mix-model significant results with percent of colonization data.
Supplementary File 5: qPCR host expression data.
Supplementary File 6: AMP resistance assay. Sheet 1: AMP sequence information. Sheet 2: Growth data parameters. Sheet 3: AMP sensitivity score.
Supplementary File 7: Genome assembly statistics for the bacterial strains sequenced in this study.
References
- 1.Hooper, L. V. Do symbiotic bacteria subvert host immunity? Nat. Rev. Microbiol. 7, 367–374 (2009).
- 2.McFall-Ngai, M. et al. Animals in a bacterial world, a new imperative for the life sciences. Proc. Natl. Acad. Sci. 110, 3229–3236 (2013).
- 3.Lloyd-Price, J. et al. Strains, functions and dynamics in the expanded Human Microbiome Project. Nature 550, 61–66 (2017).
- 4.Cabirol, A., Moriano-Gutierrez, S. & Engel, P. Neuroactive metabolites modulated by the gut microbiota in honey bees. Mol. Microbiol. 122, 284–293 (2023).
- 5.Sampson, T. R. & Mazmanian, S. K. Control of brain development, function, and behavior by the microbiome. Cell Host Microbe 17, 565–576 (2015).
- 6.Pocheron, A. L. et al. Maternal Microbiota Transfer Programs Offspring Eating Behavior. Front. Microbiol. 12, (2021).
- 7.Essock-Burns, T. et al. Maturation state of colonization sites promotes symbiotic resiliency in the Euprymna scolopes-Vibrio fischeri partnership. Microbiome 11, 1–16 (2023).
- 8.Mallott, E. K. & Amato, K. R. Host specificity of the gut microbiome. Nat. Rev. Microbiol. 19, 639–653 (2021).
- 9.Mazel, F. et al. Is Host Filtering the Main Driver of Phylosymbiosis across the Tree of Life? mSystems 3, (2018).
- 10.Sprockett, D. D. et al. Home-site advantage for host species-specific gut microbiota. Sci. Adv. 9, 1–10 (2023).
- 11.Mazel, F., Prasad, A. & Engel, P. Host specificity of gut microbiota associated with social bees: patterns and processes. Microbiol. Mol. Biol. Rev. e00080-23 (2025). doi:10.1128/mmbr.00080-23
- 12.Oliphant, K., Parreira, V. R., Cochrane, K. & Allen-Vercoe, E. Drivers of human gut microbial community assembly: coadaptation, determinism and stochasticity. ISME J. 13, 3080–3092 (2019).
- 13.Levy, R. & Borenstein, E. Metabolic modeling of species interaction in the human microbiome elucidates community-level assembly rules. Proc. Natl. Acad. Sci. U. S. A. 110, 12804–12809 (2013).
- 14.Seedorf, H. et al. Bacteria from diverse habitats colonize and compete in the mouse gut. Cell 159, 253–266 (2014).
- 15.Ganesan, R., Wierz, J. C., Kaltenpoth, M. & Flórez, L. V. How It All Begins: Bacterial Factors Mediating the Colonization of Invertebrate Hosts by Beneficial Symbionts. Microbiol. Mol. Biol. Rev. 86, (2022).
- 16.Brochet, S., Quinn, A., Mars, R. A. T. & Neuschwander, N. Niche partitioning facilitates coexistence of closely related honey bee gut bacteria. Elife 10:e68583 (2021).
- 17.Quinn, A. et al. Host-derived organic acids enable gut colonization of the honey bee symbiont Snodgrassella alvi. Nat. Microbiol. 9, 477–489 (2024).
- 18.Rothschild, D. et al. Environment dominates over host genetics in shaping human gut microbiota. Nature 555, 210–215 (2018).
- 19.Lopez, A. O., Vega, N. M. & Gore, J. Interspecies bacterial competition regulates community assembly in the C. elegans intestine. ISME J. 7, 2131–2145 (2021).
- 20.Lachat, J. et al. Hundreds of antimicrobial peptides create a selective barrier for insect gut symbionts. Proc. Natl. Acad. Sci. 121, e2401802121 (2024).
- 21.Schäfer, M. et al. Metabolic interaction models recapitulate leaf microbiota ecology. Science. 381, 42–55 (2023).
- 22.Bongrand, C. et al. Using colonization assays and comparative genomics to discover symbiosis behaviors and factors in vibrio fischeri. MBio 11, (2020).
- 23.Shetty, S. A. et al. Inter-species Metabolic Interactions in an In-vitro Minimal Human Gut Microbiome of Core Bacteria. npj Biofilms Microbiomes 8, (2022).
- 24.Culp, E. J. & Goodman, A. L. Cross-feeding in the gut microbiome: Ecology and mechanisms. Cell Host Microbe 31, 485–499 (2023).
- 25.Yu, X., Polz, M. F. & Alm, E. J. Interactions in self-assembled microbial communities saturate with diversity. ISME J. 13, 1602–1617 (2019).
- 26.Itoh, H. et al. Host–symbiont specificity determined by microbe–microbe competition in an insect gut. Proc. Natl. Acad. Sci. U. S. A. 116, 22673–22682 (2019).
- 27.Ludington, W. B. The importance of host physical niches for the stability of gut microbiome composition. Philos. Trans. R. Soc. B Biol. Sci. 379, (2024).
- 28.Bright, M. & Bulgheresi, S. A complex journey: transmission of microbial symbionts. Nat. Rev. Microbiol. 8, 218–30 (2010).
- 29.Chhun, A. et al. An engineered bacterial symbiont allows noninvasive biosensing of the honey bee gut environment. PLoS Biol. 22, 1–23 (2024).
- 30.Kwong, W. K., Engel, P., Koch, H. & Moran, N. A. Genomics and host specialization of honey bee and bumble bee gut symbionts. Proc. Natl. Acad. Sci. U. S. A. 111, 11509–11514 (2014).
- 31.Bonilla-Rosso, G. & Engel, P. Functional roles and metabolic niches in the honey bee gut microbiota. Curr. Opin. Microbiol. 43, 69–76 (2018).
- 32.Ellegaard, K. M. et al. Genomic changes underlying host specialization in the bee gut symbiont Lactobacillus Firm5. Mol. Ecol. 28, 2224–2237 (2019).
- 33.Prasad, A. et al. Evolution of gut microbiota across honeybee species revealed by comparative metagenomics. Nat. Commun. 16, 9069 (2025).
- 34.Kwong, W. K. & Moran, N. A. Gut microbial communities of social bees. Nat. Rev. Microbiol. 14, 374–384 (2016).
- 35.Cabirol, A. et al. Fecal transplant allows transmission of the gut microbiota in honey bees. mSphere 9, e00262-24 (2024).
- 36.Powell, J. E., Martinson, V. G., Urban-mead, K. & Moran, A. Routes of acquisition of the gut microbiota of the Honey Bee Apis mellifera. Appl. Environ. Microbiol. 80, 7378–7387 (2014).
- 37.Horak, R. D., Leonard, S. P. & Moran, N. A. Symbionts shape host innate immunity in honeybees: Symbionts shape honey bee immunity. Proc. R. Soc. B Biol. Sci. 287, (2020).
- 38.Kwong, W. K., Mancenido, A. L. & Moran, N. A. Immune system stimulation by the native gut microbiota of honey bees. R. Soc. Open Sci. 4, 1–9 (2017).
- 39.Guo, L., Tang, J., Tang, M., Luo, S. & Zhou, X. Reactive oxygen species are regulated by immune deficiency and Toll pathways in determining the host specificity of honeybee gut bacteria. Proc. Natl. Acad. Sci. 120, e2219634120 (2023).
- 40.Kwong, W. K. et al. Dynamic microbiome evolution in social bees. Sci. Adv. 3, 1–17 (2017).
- 41.Ellegaard, K. M., Suenami, S., Miyazaki, R. & Engel, P. Vast Differences in Strain-Level Diversity in the Gut Microbiota of Two Closely Related Honey Bee Species. Curr. Biol. 30, 2520-2531.e7 (2020).
- 42.Li, Y., Leonard, S. P., Powell, J. E. & Moran, N. A. Species divergence in gut-restricted bacteria of social bees. Proc. Natl. Acad. Sci. U. S. A. 119, 1–10 (2022).
- 43.Kešnerová, L. et al. Gut microbiota structure differs between honeybees in winter and summer. ISME J. 14, 801–814 (2020).
- 44.Andrews, S., Krueger, F., Seconds-Pichon, A., Biggins, F. & Wingett, S. FastQC. A quality control tool for high throughput sequence data. Babraham Bioinformatics. Babraham Institute (2015).
- 45.Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114–20 (2014).
- 46.Bankevich, A. et al. SPAdes: A new genome assembly algorithm and its applications to single-cell sequencing. J. Comput. Biol. 19, 455–477 (2012).
- 47.Parks, D. H., Imelfort, M., Skennerton, C. T., Hugenholtz, P. & Tyson, G. W. CheckM: assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes. Genome Res. 25, 1043–1055 (2015).
- 48.Seemann, T. Prokka: Rapid prokaryotic genome annotation. Bioinformatics 30, 2068–2069 (2014).
- 49.Hyatt, D. et al. Prodigal: prokaryotic gene recognition and translation initiation site identification. BMC Bioinformatics 11, 119 (2010).
- 50.Shaffer, M. et al. DRAM for distilling microbial metabolism to automate the curation of microbiome function. Nucleic Acids Res. 48, 8883–8900 (2020).
- 51.Emms, D. M. & Kelly, S. STAG: Species Tree Inference from All Genes. bioRxiv 1–29 (2018).
- 52.Emms, D. M. & Kelly, S. OrthoFinder: Phylogenetic orthology inference for comparative genomics. Genome Biol. 20, 1–14 (2019).
- 53.Segura Munoz, R. R. et al. Experimental evaluation of ecological principles to understand and modulate the outcome of bacterial strain competition in gut microbiomes. ISME J. 16, 1594–1604 (2022).
- 54.Ho, P. Y., Nguyen, T. H., Sanchez, J. M., DeFelice, B. C. & Huang, K. C. Resource competition predicts assembly of gut bacterial communities in vitro. Nature Microbiology 9, (2024).
- 55.Aranda-Díaz, A. et al. Assembly of stool-derived bacterial communities follows “early-bird” resource utilization dynamics. Cell Syst. 16, 101240 (2025).
- 56.Zheng, H., Powell, J. E., Steele, M. I., Dietrich, C. & Moran, N. A. Honeybee gut microbiota promotes host weight gain via bacterial metabolism and hormonal signaling. Proc. Natl. Acad. Sci. 114, 4775–4780 (2017).
- 57.Engel, P., Martinson, V. G. & Moran, N. A. Functional diversity within the simple gut microbiota of the honey bee. Proc. Natl. Acad. Sci. U. S. A. 109, 11002–11007 (2012).
- 58.Ellegaard, K. M. & Engel, P. Genomic diversity landscape of the honey bee gut microbiota. Nat. Commun. 10, (2019).
- 59.Lang, H. et al. Specific Strains of Honeybee Gut Lactobacillus Stimulate Host Immune System to Protect against Pathogenic Hafnia alvei . Microbiol. Spectr. 10, (2022).
- 60.Emery, O., Schmidt, K. & Engel, P. Immune system stimulation by the gut symbiont Frischella perrara in the honey bee (Apis mellifera). Mol. Ecol. 26, 2576–2590 (2017).
- 61.Raymann, K. & Moran, N. A. The role of the gut microbiome in health and disease of adult honey bee workers. Curr. Opin. Insect Sci. 26, 97–104 (2018).
- 62.Moriano-Gutierrez, S. et al. Critical symbiont signals drive both local and systemic changes in diel and developmental host gene expression. Proc. Natl. Acad. Sci. U. S. A. (2019).
- 63.Lamarcq, L. H. & Mcfall-Ngai, M. J. Induction of a Gradual , Reversible Morphogenesis of Its Host ’ s Epithelial Brush Border by Vibrio fischeri. Infect. Immun. 66, 777–785 (1998).
- 64.Koch, E., Moriano-Gutierrez, S., Ruby, E., McFall-Ngai, M. & Liebeke, M. The impact of persistent colonization by Vibrio fischeri on the metabolome of the host squid Euprymna scolopes. J. Exp. Biol. jeb.212860 (2020).
- 65.Martino, M. E. et al. Bacterial Adaptation to the Host’s Diet Is a Key Evolutionary Force Shaping Drosophila-Lactobacillus Symbiosis. Cell Host Microbe 24, 109-119.e6 (2018).
- 66.Liberti, J. et al. Gut microbiota influences onset of foraging-related behavior but not physiological hallmarks of division of labor in honeybees. MBio 15, (2024).
- 67.Koch, E. J., Miyashiro, T., McFall-Ngai, M. J. & Ruby, E. G. Features governing symbiont persistence in the squid-vibrio association. Mol. Ecol. 23, 1624–1634 (2014).
- 68.Wu, J. et al. Honey bee genetics shape the strain-level structure of gut microbiota in social transmission. Microbiome 9, 225 (2021).
- 69.Casteels, P., Ampe, C., Jacobs, F., Vaeck, M. & Tempst, P. Apidaecins: Antibacterial peptides from honeybees. EMBO J. 8, 2387–2391 (1989).
- 70.Franzenburg, S. et al. Distinct antimicrobial peptide expression determines host species-specific bacterial associations. Proc. Natl. Acad. Sci. U. S. A. 110, (2013).
- 71.Thaiss, C. A. et al. Microbiota Diurnal Rhythmicity Programs Host Transcriptome Oscillations. Cell 167, 1495-1510.e12 (2016).
- 72.Chu, H. & Mazmanian, S. K. Innate immune recognition of the microbiota promotes host-microbial symbiosis. Nat. Immunol. 14, 668–675 (2013).
- 73.Login, F. H. et al. Antimicrobial Peptides Keep Insect Endosymbionts Under Control. Science. 334, 362–365 (2011).
- 74.Lazzaro, B. P., Zasloff, M. & Rolff, J. Antimicrobial peptides: Application informed by evolution. Science. 368, (2020).
- 75.Rojasa, A. A.-, Frahm, D., Hurwitz, R., Brinkmann, V. & Iatsenko, I. Resistance to host antimicrobial peptides mediates resilience of gut commensals during infection and aging in Drosophila. Proc. Natl. Acad. Sci. 120, e2305649120 (2023).
- 76.Marra, A., Hanson, M. A., Kondo, S., Erkosar, B. & Lemaitre, B. Drosophila Antimicrobial Peptides and Lysozymes Regulate Gut Microbiota Composition and Abundance. MBio 12, (2021).
- 77.Belkaid, Y. & Hand, T. W. Role of the microbiota in immunity and inflammation. Cell 157, 121–41 (2014).
- 78.Baishya, J. & Wakeman, C. A. Selective pressures during chronic infection drive microbial competition and cooperation. npj Biofilms Microbiomes 5, (2019).
- 79.Ye, Z. et al. Synergistic collaboration between AMPs and non-direct antimicrobial cationic peptides. Nat. Commun. 15, (2024).
- 80.Ha, E. M., Oh, C. T., Bae, Y. S. & Lee, W. J. A direct role for dual oxidase in Drosophila gut immunity. Science. 310, 847–850 (2005).
- 81.Ha, E. M. et al. An antioxidant system required for host protection against gut infection in Drosophila. Dev. Cell 8, 125–132 (2005).