Investigating Aerobic Hive Microflora: Role of Surface Microbiome of Apis mellifera

2. Methods

The standard media employed (GRM, GMF, Czapek, Sabouraud, Soy, A4, Gauze II; see Supplementary Table 1) comprised both liquid and solid (20% agar) preparations. The bees (collected on 01/12/2021 from three hives) were stored individually at -20°C. To isolate cuticular microbes, five bees were held for one hour in sterile containers (150 mL), rinsed with 1 mL of water, and plated directly or after pre-enrichment in Czapek broth. Pure cultures were obtained by sequentially processing body parts (300-1200 µL wash/media), vortexing (1 min), and inoculating onto selective media (2–10 days incubation). Following this, enriched samples were plated in order to isolate pure lines. Further experiments were then conducted.

All obtained isolates were cultured on appropriate solid media. Their morphology and cellular structure were determined in unstained preparations and, if necessary, stained with methylene blue or Lugol solution, Gram and Zill-Nelsen stains. If all parameters coincided in isolates from the same source, they were considered identical and then only one of the lines was cultured, preserved and analyzed. In the article, the names of the crops correspond to their Supplementary photos.

For experiments on “shedding” from the cuticle surface, inoculations were performed from initially sterile surfaces with which the bees were in contact. 5 bees were kept for one hour in sterile 150 ml containers. The containers were then rinsed with 1 ml of water, which was transferred to solid medium directly or first enriched in 10 ml of Czapek's nutrient medium.

A series of experiments were performed to select the optimum conditions for the development in the elective medium and to select techniques for flushing microbiomes. In the first one, after adding 1200 µl of water to the sample, the sample was vortexed for one minute, after which the flush and the sample itself were placed into a series of elective media (10 ml each) with varying pH from 6 to 8 in 0.2 increments. Every 24 hours, 1 ml of the elective medium was transferred to solid media for further separation of pure lines. Three different compositions of washing solutions were used: (1) distilled water; (2) 0.9% NaCl in water; (3) 0.001% SDS, 0.9% NaCl in water.

For extraction from the surface of the bees, the limbs were carefully separated from the body; the cuticle was incised from above and carefully separated. To avoid contamination, the lower part of the head and the last two segments of the abdomen were not used.

To determine the differences between the microbiota of the different parts of the surfaces, bees were cut with sterile needles, the intestines were removed and their microbiota were analyzed separately to exclude contamination: identical lines from the surface and from the intestines were not considered. Samples were placed in 300 μl of 0.9% NaCl solution, vortexed for one minute, and placed in 10 ml of a series of elective media with and without antimycotics added. The media were placed on a rocker for a week, from day 3 to day 15, 200 µl of media were transferred to solid media every other day to obtain isolates.

Tests for the production of antimicrobial compounds were performed. For the dot test, the test culture was transferred to the center of the petri dish and the inhibiting culture was transferred in 4 symmetrical dots at a distance of 2.5 cm from the center. For the line test, the suppressed culture was transferred along the diameter with a continuous stroke using a 2 mm thick glass rod. Perpendicular to it, the inhibitory culture was applied with 4-5 parallel strokes with a 2 mm glass rod. The size of the suppression zone was defined as the minimum distance between inhibiting and suppressed cultures. For CFU count and suppression zone measurement ImageJ was used [19].

For metagenomics and 16S quantifications 10 bees were collected from three hives. After washing, samples were enriched on Czapek nutrient medium for 3 days and then sequenced. DNA isolation from isolates was performed using the DNeasy Blood & Tissue kit (QIAGEN, Hilden, Germany) according to the protocol recommended by the manufacturer. DNA library was quantified and quality-assured with a capillary electrophoresis TapeStation 4200 (Agilent, Santa Clara, CA, USA). For 16S rRNA sequencing we use Quick-16S NGS Library Prep Kit (Zymo Research, Irvine, CA, USA). 16S sequencing rRNA sequencing was performed using Illumina MiSeq (San Diego, CA, USA), expected read length 250 bp.WGS sequencing was performed using the Illumina HiSeq (San Diego, CA, USA), expected read length 100 bp. Data available at BioProject PRJNA1048732.

16S amplicon sequencing data analysis was performed using QIIME 2 [20] (version 2024.5.0). Following import, the DADA2 module [21] was applied for data denoising. Taxonomy was assigned using naive Bayes classifiers with the SILVA [22], GreenGenes [23], and BEExact [24] databases. A phylogenetic tree was constructed using the IQ-TREE [25] plugin (version 2.3.4). Metabolic annotation was carried out using the PICRUSt2 [26] plugin (version 2.5.3).

WGS data trimming was performed using fastp [27] (version 0.23.4) with a minimum quality threshold of Q20, a preference for front trimming, and a sliding window of size 10. For metagenomic annotation, comprehensive analysis was conducted using Kraken 2 [28] (version 2.1.3) with Bracken [29] (version 2.9) correction on the pre-built NCBI nucleotide database (nt). To annotate potential fungal metagenomic sequences, Kaiju [30] (version 1.10.1) was employed with the pre-built RefSeq fungal sequences index.

To eliminate possible contamination, we compared our data with sequencing data from the same hive’s combs [31] (BioProject PRJNA1048732) and with pan-metagenomic data [9] (BioProject PRJNA879967). Host and human reads were removed using Bowtie 2 [32] (version 2.5.2) with the ‘very-sensitive-local’ preset. Genomes for dehosting and decontamination are available in NCBI GenBank under the following IDs: GCA_003254395.2, GCA_029169275.1, GCA_000184785.2, GCA_000469605.1, GCA_014066325.1, GCA_000001405.29.

Metagenome assembly was performed using metaSPAdes [33] and IDBA-UD [34] (version 1.1.3). After quality control with MetaQUAST [35] (version 5.2.0), IDBA-UD was selected with a k-mer step of 20 and k-mer sizes ranging from 16 to 116 which produced maximal N50. The PROKKA [36] tool (version 1.14.5) was then used to perform metabolism reconstruction on the selected assembly.

For data visualization we use R [37] with libraries tidyverse [38], vegan [39], ape [40] and aRchiteutis [41].

3. Results

The cultivation conditions and media were optimised through experimentation utilising a range of sample sources.

Keeping bees in sterile medium and further sowing the flushes onto solid media yields surprisingly low CFU counts (Figure 1a1), no more than 20 in five experiments. Sowing the flushes directly onto nutrient media leads to the detection of relatively small CFU of fungi in the medium (Figure 1a2), suppressing the activity of other microorganisms, resulting in no real cell counts in the medium. The highest biodiversity is achieved after pre-cultivation on enrichment liquid medium (Figure S1).

To account for organisms not densely attached to the surface of bees, the microbiota obtained from the walls of the container in which the bees were transported were analyzed. Only Aspergillus and Penicillium were detected on Soburo's medium after direct bee washing; other species were cultured on Czapek's medium. The composition of organisms obtained on Czapek's, GRM and GMF media did not differ.

Seeding from abdomen, limbs and cuticular hairs was also performed to optimize the technique (Figure S3). When using a solution with a mixture of compounds to disrupt biofilms, cultures could not be isolated (Figure S2c). The highest diversity was obtained in cultures on solid nutrient agar after enrichment on liquid nutrient medium for 24 hours, isolating 100-5000 CFU from different bee organs (Figure S3).

Different cultures with complex morphology, which was often not preserved during passages, as well as mixed colonies consisting of cells of different morphologies and probably from several taxa were frequently observed in the cultures (Figure S4). In total, more than 250 colonies were obtained from the cuticle of 10 adult worker winter bees, of which 54 differed in morphological and physiological criteria (Figure S5-13). Of these, only a small fraction (not more than 20%) are also readily isolated from the hive environment. Cultures S5c, S5d and S6d (Aspergillus spp.) and S6a (Penicillium sp.) dominated the cultures from dry dead bees from the same hive.

2 different cultures were obtained that inhibited the development of other microorganisms (Figure 1b). Cultures similar to S10b were independently isolated from 4 out of 10 bees tested from 2 out of 3 different hives. Culture S10f was isolated only once from a sample in which S10b was also present.

Only the culture of the unknown Actinomycete S10b retained antimicrobial activity when transferred to other media. The highest zone of suppression was observed on Czapek medium, while on A4, Soy or Gauze II agar the activity was lost. For further studies, tests were performed after sowing with dots (Figure 1c, Figure S14) or lines (Figure 1d, Figures S14-15). Some cultures are more suppressed than others (Figure S16). On average, activity against fungi is more strongly expressed than antibacterial activity (ANOVA p-value = 0.003), however, other obtained fungi producing antimicrobial compounds were not suppressed (Figure 1d6). It is also possible that cultures obtained from the surface of bees are more resistant to the antimicrobial compounds produced (ANOVA p-value < 10^-4). However, more testing is needed to draw definite conclusions about this.

The pooled microbiota sample was analyzed by 16S and WGS sequencing methods (Figure 2). The samples show an atypically high proportion of unclassifiable sequences for 16S sequencing, which may indicate the presence of unknown species or possible sample preparation problems. Different databases were used for annotation, all of which give similar results when used in the analysis and do not help to solve the low classification rate problem (Figure S17).

Alpha diversity indices are similar for indices calculated from 16S sequencing data of V1-V2, V3-V4 and WGS: Shannon index 1.59±0.05, 1.41±0.18 and 1.63, Simpson index 0.70±0.01, 0.68±0.06 and 0.70. respectively. Alpha diversity is similar to that obtained in other articles (ANOVA p-value 0.13) (Figure S18). For beta diversity, the annotation results differ from the rest of the pan-metagenome. This is due to sample preparation, especially - enrichment of the sample.

Metagenome annotation identified 23955 CDSs, of which 10588 were able to identify product function. Thirty-seven genes (26 unique) for biosynthesis of 22 different antimicrobial compounds (Supplementary table S2) and 119 resistance genes (40 unique) were identified. Complete toxin-antitoxin systems higA/B, mqsA/R were detected.

Antimicrobial compounds possibly produced by the metagenome can vary in origin, mechanism of action, and chemical nature; they may be small molecules or antimicrobial peptides, such as bacitracin and gramicidin. Each antimicrobial compound can possess either a narrow mechanism of action or a broad spectrum of antimicrobial activity.

Copper, cobalt-zinc-cadmium, and arsenic resistance cassettes were also detected. Several organic hydroperoxide resistance genes were found: ohrR (5), ohrB (3), ohrA (2). The community is characterized by a large spectrum of excreted and metabolizable secondary metabolites.

4. Discussion

The bee cuticle community grows on a relatively poor substrate. Studies with whole bee 16S sequencing in the methodology are encountered, but show little or no difference with gut sequencing [42]. Small values are shown by qPCR analysis and culture methods [9,10,43]. The bee surface microbiome is probably poor, with no more than a few thousand CFUs there. The highest biomass is probably characteristic of mold fungi (in contrast with previous research), while the maximum biodiversity is probably among bacteria and yeasts getting there from sugar-rich nectar and early honey (similar to previous studies [9,43]).

In the transition to opportunism or parasitism of individual members of the community, such an environment becomes richer for a short time. Community succession is complete after bee death, the final community is probably dominated by the mold fungi Penicillium and Aspergillus. The extent to which succession is sequential is unclear. It is also unclear what role the fungi, which are potential but not obligate pathogens [44,45], play in the cuticular microflora.

Metagenome annotation revealed more than 20 000 coding sequences, half of them linked to specific functions. A significant portion was associated with the biosynthesis of various antimicrobial compounds, while numerous resistance genes were also identified. Some members of the community produce antimicrobial compounds. Whole-genome sequencing also confirms the presence of genes for their biosynthesis and resistance. The key producers of antimicrobial compounds may be Burkholderiaceae, Bacilli and Actinobacteria. The detected components of toxin-antitoxin systems may also be involved as part of the community resistance maintenance system.

Fungicides and insecticides are known to affect this community [11]. It remains unclear to what extent they will affect its bark representatives, given the diversity of resistance of organisms from the surface.

The results of the metagenomic analysis differ from the known literature data on bee microbiology [2,9,31,46,47]. Previously, 16S sequencing of bee cuticle was performed [9], which was reannotated in this work (Figure 3). The community is dominated by impurities from intestinal microflora: obligately anaerobic Lactobacillus, Bombilactobacillus, Snodgrassella. The share in the microbiome of aerobes characteristic of the foregut and other hive environments - Apilactobacillus and Actinobacteria, probably represented by Bifidobacterium asteroides - is high. In our study, after enrichment, we find a different situation, but it is difficult to say which approach better describes the real, influential bee cuticular community.

The alpha diversity indices from 16S and whole genome sequencing are comparable, consistent with all studies. However, beta diversity results varied across the pan-metagenome. Despite some similarities, there are also obvious differences with the intestinal microflora [2,15,48]. Unclassifiable species are more widely represented in the cuticle community and may play an important role in it. Key isolated actinomycete cultures remain unclassifiable by both culture and metagenomic methods. Their further study in confirming their role and persistence in bee cuticle communities is very important.

5. Conclusions

Our work is one of the first studies of the bee cuticular community. Despite the application of various methods (16S sequencing, WGS, qPCR, cultural methods) the question still remains: what is the functionality of this community? What are the introduced species in it, and does it in principle have its own bark species? Our study supports the hypothesis that this community is independent, but unequivocal confirmation may come from analysis of the surface microflora by FISH or SEM. Metagenomic analyses and qPCR are hampered by significant contamination of the bee surface by representatives of their gut microbiome, which are obligate anaerobes. There is probably a higher proportion of actinomycetes and unclassifiable organisms in the community than in the bee gut. Their source may be the honeycomb of the hive, but the opposite situation may also be true, and finally different actinomycetes and unclassifiable organisms may be present on the surface of the hive and bees. This question requires further research.

Our results open a new chapter in micro”bee”ology. Currently, there are no fully unstudied hive environments from a metagenomic perspective. The beehive, functioning as a superorganism, incorporates the genomes of its associated organisms into a unified hologenome, collectively responding to external stimuli.