Defensive symbionts of the European beewolf face competition from brood cell microbiota during vertical transmission

Abstract

Beewolf wasps rely on an ancient defensive symbiosis with Streptomyces bacteria that protect their larvae from fungal infection. Female beewolves apply the bacteria to the brood-cell ceiling, and larvae later transfer the symbionts onto the cocoon surface, where they produce antifungal metabolites. Here, we investigated the mechanism of symbiont transfer from the beewolf brood cell to the larval cocoon and characterized the microbial community dynamics across different beewolf life stages and during larval hibernation. Fluorescence in situ hybridization revealed that the symbionts are transiently taken up into the proximal midgut lumen and then regurgitated onto the cocoon during the spinning process. High-throughput sequencing showed that the bacterial community of beewolf feeding larvae resembles that of the honeybee prey, whereas that of adults and diapausing larvae is dominated by Wolbachia. Moreover, the cocoon bacterial community is initially dominated by the defensive Streptomyces philanthi symbiont, but when larvae excrete the gut content inside the cocoon, other bacterial taxa including Lactobacillus, Gilliamella, and Bartonella shift the community composition toward dominance by Pseudomonadota. Our findings provide new insights into the transmission route of an ancient extracellular symbiont and its potential competition with other bacteria in this long-term defensive symbiosis.

Defensive Streptomyces symbionts of beewolf wasps experience a complex vertical transmission route from adult antennae via larval guts to the cocoon, which entails interactions with the gut microbiota and environmental microbes.

Introduction

Insects are among the most abundant and speciose multicellular eukaryotes on Earth. Their evolutionary success relies on evolutionary adaptations that enable them to thrive in diverse ecological niches. Many insects obtain novel metabolic capabilities including nutrient supplementation, digestion, detoxification, antimicrobial defense, and environmental stress tolerance by engaging in symbiotic associations with microbes (Kaltenpoth et al. 2025). The stability and fidelity of symbiotic associations fundamentally depend on the modes of transmission of the microbial partners across host generations (Bright and Bulgheresi 2010). Three principle routes of transmission exist, each one having specific implications for the evolution of the symbiosis; horizontal, vertical, and mixed-mode transmission (Salem et al. 2015).

In associations with vertical transmission, the symbionts are inherited from parent to offspring, resulting in the alignment of host’s and symbiont’s evolutionary interests and therefore selecting for symbiont traits that enhance host fitness (Bright and Bulgheresi 2010). Vertically transmitted symbionts are often maintained in specialized organs that provide an optimized nutritional environment to the symbionts and protects them from the host’s immune system as well as against antagonists and competitors (Ferrarini et al. 2022). The tight and continuous association with the host results in genomic streamlining and metabolic specialisation of the symbiont, enhancing cooperative efficiency and reducing metabolic redundancy in the system (Mccutcheon and Moran 2012). Vertical transmission occurs transovarially for most intracellular symbionts (Szklarzewicz and Michalik 2017), whereas extracellular symbionts can be transmitted via a diversity of mechanisms, including egg smearing, specialized secretions or capsules adjacent to the eggs, or by applying the symbionts to the direct environment of the developing offspring (Salem et al. 2015). All of the extracellular transmission routes have in common that the symbionts are exposed to the external environment for—short to very long—periods, where they may have to compete with free-living microorganisms. However, little is known about how prevalent competition is in extracellularly transmitted symbioses, and how the symbionts prevail to successfully colonize the new host generation.

Beewolves (Hymenoptera and Crabronidae), a group of solitary digger wasps, live in a symbiotic association with Streptomyces philanthi bacteria that produce antibiotics to defend the beewolf larvae against pathogenic fungi during hibernation in the soil (Kaltenpoth et al. 2005). The symbionts are cultivated in specialized antennal gland reservoirs (AGR) of females (Goettler et al. 2007, 2022) and secreted to the ceiling of the subterranean larval brood cells in a hydrocarbon-rich matrix prior to oviposition (Kaltenpoth et al. 2005). The hydrocarbon-rich secretion serves as physicochemical barrier that protects symbionts against high amounts of toxic nitric oxide produced by beewolf eggs (Ingham et al. 2023a, b). This production of nitric oxide is a specialized extension of the immune response of the beewolf egg that effectively sanitizes the brood cell (Strohm et al. 2019). Later, the symbionts are taken up by the larvae and incorporated into the silken cocoon during the process of cocoon-spinning. Antibiotics are produced by the symbiont on the cocoon within the first 2 weeks after cocoon spinning, and the compounds remain stable throughout the entire beewolf hibernation period, protecting the larva against pathogenic fungi (Kroiss et al. 2010, Koehler et al. 2013, Koehler and Kaltenpoth 2013, Engl et al. 2018). After periods of hibernation of up to 9 months, beewolves undergo metamorphosis, and females acquire the symbionts from the cocoon shortly before eclosion (Fig. 1), whereas adult males eclose symbiont-free. The mechanism of symbiont acquisition is not completely elucidated, but the Streptomyces cells from the cocoon are likely transferred to the AGRs when females rub the antennae against the internal surface of the cocoon prior to eclosion (Kaltenpoth et al. 2009). Despite this elaborate vertical transmission route, horizontal transmission seems to occur occasionally, as suggested by incongruences between host and symbiont phylogenies (Kaltenpoth et al. 2014).

Figure 1.

Beewolf life cycle and symbiont transmission. (a) Beewolf females harbor symbiotic Streptomyces bacteria in specialized AGRs. Prior to or during female eclosion from the cocoon, the AGRs are colonized by Streptomyces cells from the cocoon. (b) Prior to oviposition, beewolf females apply antennal gland secretion containing S. philanthi (Sph) to the ceiling of the brood cell. (c) 6–8 days after hatching from the egg, the beewolf larva spins a cocoon while taking up the secretion with its mouthparts and incorporating symbiont cells onto the cocoon surface. (d) Larvae release their gut content 3–4 days after cocoon spinning and enter diapause. S. philanthi produces a cocktail of antibiotics that effectively protects the larvae against detrimental fungi during hibernation. (e) Samples for microbial community profiling included in this study.

Considering the harsh environment and prolonged times that symbiont cells must prevail on cocoon fibers for emerging females to acquire them and colonize their AGRs, it is intriguing how specificity of transmission and stability across generations is maintained in the system. This is especially interesting considering the high potential for competition with environmental (see Engl et al. 2016) and beewolf gut-associated microbes in the brood cell. Indeed, filamentous Actinobacteria from the genera Streptomcyes, Amycolatopsis, and Nocardia have occasionally been found growing in high density and likely replacing the symbiont within female antennae (Nechitaylo et al. 2014). Consequently, symbiont replacement in aposymbiotic beewolf females was successfully conducted by in vitro infection assays with an Amycolaptosis strain that was previously obtained from beewolf antennae (Kaltenpoth et al. 2014). Moreover, Streptomyces symbionts likely encounter the beewolf larval gut microbes during oral symbiont acquisition from the brood cell, as well as on the cocoon, when the larva excretes its gut content prior to hibernation. However, the localization of the symbionts in the short period between oral acquisition and secretion into the brood cell, as well as the potential for competition with other microbes in the larval gut and on the cocoon remain unknown.

To address these knowledge gaps, we conducted 16S rRNA amplicon sequencing to characterize the gut microbial communities of beewolves across different life stages and on cocoons. Moreover, we investigate the transmission of symbionts in beewolves from the brood cell to the cocoon by a combination of microscopy-based and molecular techniques. Our results indicate that larvae transiently store the Streptomyces symbionts in the foregut after acquisition from the brood cell, and then regurgitate them onto the cocoon during the spinning process. Both in the larval gut and on the cocoon, the symbionts have to survive the encounter with gut bacteria and environmental microbes. Despite a decrease in relative abundances of the symbionts on the cocoon during the long period of hibernation, part of the population survives and ensures transmission to the next beewolf generation.

Materials and methods

Beewolf rearing

European beewolf females (Philanthus triangulum) were collected in Berlin and Mainz, Germany. They were placed in observation cages (Strohm and Linsenmair 1995) in a greenhouse [14 h day, 10 h night; 23°C (+/− 3°C fluctuation)] and provided with honeybees (Apis melifera) and honey ad libitum. Beewolf females were monitored as they constructed underground nests that were accessible for obtaining larvae or adults at defined time points.

Preparation of beewolf larvae for histology and fluorescence in situ hybridization

For the morphological assessment and the localization of the symbiont S. philanthi via fluorescence in situ hybridization (FISH), the following larval samples were collected: Four larvae after uptake of the symbiont-containing white substance before spinning the cocoon, two larvae 2 days after cocoon spinning, and two larvae 5 days after cocoon spinning. These time points were chosen to compare the occurrence and distribution of bacteria in the gut of European beewolf larvae before and after gut content release onto the cocoon, which happens 3–4 days after cocoon spinning.

The larvae were fixed in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) for 24 h, then transferred to chloroform, where they remained for another 24 h. Larvae were divided into anterior and posterior segments and postfixed in 4% PFA for 24 h. Afterwards, the samples were washed in running tap water for 30 min and dehydrated in an ascending series of n-butanol. This was followed by embedding in Technovit 8100 (Kulzer, Wehrheim, Germany; Weiss and Kaltenpoth 2016). From each larva, two transversal semithin section series (8 µm) were generated on a rotary microtome (Leica RM 2245) with glass blades, by distributing the sections alternately to two microscope slides. The sections were transferred to silanized glass slides in water drops and dried for 30 min at 50°C. One of the two series was stained with haematoxylin and eosin (HE) and used for morphological evaluation, the other was used for FISH.

Histological evaluation of symbiont localization

Morphological evaluation was performed on stained sections of P. triangulum larvae. The beewolf cell nuclei were stained with Mayer’s haematoxylin (Roth, Karlsruhe, Germany) for 10 min, followed by two rinses in cold tap water and a bluing step in lukewarm tap water for 10 min. The cytoplasm staining was performed with eosin (0.5% aqueous) for 4 min. After two washing steps with cold tap water, a short differentiation step in 96% ethanol followed. Afterwards, the sections were briefly rinsed with distilled water and dried for 30 min at 50°C, before they were mounted with Histokitt (Roth). The preparations were assessed on an AxioImager.Z2 microscope (Carl Zeiss, Jena, Germany).

Symbiont localization by FISH

Localization of S. philanthi was performed by FISH with the specific 16S rRNA probe SPT-177 (5'-Cy3-CACCAACCATGCGATCGGTA-3') (Kaltenpoth et al. 2006, 2012) and a general probe targeting Eubacteria (EUB-388-Cy3) (Amann et al. 1990). The counterstaining of host cell nuclei was achieved with DAPI. Hybridization was performed by incubation for 90 min at 50°C in 100 µl hybridization buffer (0.9 M NaCl, 0.02 M Tris/HCl pH 8.0, 0.01% SDS) containing 5 µl of SPT-177-Cy5 (500 nM) and 5 µl of EUB-388-Cy3 (500 nM), as well as 1 µl of DAPI (5 µg/ml). Afterwards, the sections were washed for 20 min in washing buffer (0.1 M NaCl, 0.02 M Tris/HCl pH 8.0, 0.01% SDS, 5 mM EDTA) preheated to 50°C, then in distilled water for 20 min and subsequently covered with VectaShield (Vector, Burlingame, CA, USA). The preparations were evaluated on an AxioImager.Z2 fluorescence microscope with Apotome.2 (Carl Zeiss).

3D-reconstruction and symbiont localization

For the three-dimensional reconstruction, one larva after the uptake of the symbiont-containing white substance and before spinning the cocoon was used. After fixation in 4% formaldehyde and chloroform, embedding in Technovit 8100 was carried out. A transverse section series (8 µm) was created, of which every second slide (of a total of 110 slides) was used for FISH as described above. Images were obtained from the hybridized sections, which were divided into four categories depending on the symbiont abundance: 0 = absence of S. philanthi; + = few isolated cells of S. philanthi; ++ = intermediate number of S. philanthi cells, and +++ = high number of S. philanthi cells. After the examination was completed, the cover slips of the preparations were removed by placing the slides in water for about 2 h. Then all preparations (slides that were used for FISH and slides that were not stained before) were stained with HE and mounted with Histokitt. Images of all sections were acquired on an AxioImager.Z2 and converted to TIF image stacks with Fiji (Schindelin et al. 2012). Subsequently, image alignment and 3D-reconstruction were performed with Amira 6 (Fei, Hillsboro, OR, USA.) The definition of symbiont abundance categories were based on the evaluated FISH images and inserted into the reconstruction as heatmap color coding.

Symbiont localization by FISH in combination with detection by quantitative polymerase chain reaction (qPCR)

A larva after the ingestion of symbiont-containing white substance and before spinning the cocoon was fixed with Carnoy´s fixative (60% ethanol, 30% chloroform, 10% acetic acid) for 24 h. After fixation, the larva was washed several times in 96% ethanol and divided transversely into six equally sized segments. For better alignment and orientation, the segments were pre-embedded in agar (1% in distilled water), then dehydrated with n-butanol and transferred via three washes in isopropanol to paraffin wax (Roth) at 60°C. After two embedding steps in paraffin wax at 60°C for 24 h each, the samples were sectioned on a rotary microtome (Leica RM 2245) with s35 metal blades (Feather pfm, Pyonton, UK). From each segment, 1–3 sections (6 µm) were obtained, mounted on silanized slides and dried for 1 h at 50°C. After dewaxing in xylene and rehydration with a descending ethanol series, sections were used for FISH (EUB-388-Cy3, SPT-177-Cy5, DAPI) and served for the microscopic evaluation of symbiont abundance in the individual larval segments. The remaining parts of the larval segments in the paraffin blocks were trimmed out with a microtome blade, and paraffin was removed in three batches of xylene, for 3 h each. After washing five times in absolute ethanol for at least 1 h each, the DNA was extracted for qPCR (qPCR) using the Epicentre MasterPureTM DNA extraction kit (Epicenter Technologies, Madison, USA). Extraction was carried out according to the manufacturer’s instructions with the following modifications: Samples were incubated at 37°C for 1 hour in Tissue and Cell Lysis solution after adding Lysozyme to a final concentration of 8 µg/ml. An additional centrifugation step (14 000 r/m, 2 min) was applied after protein precipitation. Extracted DNA was resuspended in 50 μl low TE buffer. Symbionts on the larval sections were quantified via qPCR as described in (Ingham et al. 2023b). Briefly, the 16S rRNA of S. philanthi was amplified using the specific primers Strep_phil_fwd3mod (5ʹ TGGTTGGTGGTGGAAAGC 3ʹ) and S16S_rev (5ʹ GTGTCTCAGTCCCAGTGTG 3ʹ) (Kaltenpoth et al. 2010). PCR reactions were prepared with 5x HOT FIREPol® EvaGreen® qPCR Mix Plus (Solis BioDyne, Tartu, Estonia). Amplification was performed on a Rotor Gene Q cycler (Qiagen, Hilden, Germany). Each reaction consisted of 6 μl H2O, 2 μl EVA Green Mix, 0.5 μl of each primer, and 1 μl of the diluted DNA template. A 135 bp amplicon of the 16S rRNAwas obtained by PCR with the same primers as above, purified, and used as a template to prepare a dilution series from 10⁹ to 10⁶ copies/μL. DNA copy number of samples was then estimated from the quantification cycle (Cq) values from the standard curve.

Symbiont localization on beewolf cocoons by FISH

Three cocoons containing male beewolf larvae collected five days after spinning and one cocoon with a female beewolf larva collected two months after spinning were used for symbiont localization by FISH following previously described protocols (Weiss and Kaltenpoth 2016). For this purpose, the anterior parts of the cocoons were cut off, fixed in Carnoy and pre-embedded in Agar (1% in distilled water). The agar blocks were then embedded in paraffin and processed into sections (6 µm) on the rotary microtome as described above for the larval sections. FISH with the Probes EUB-388-Cy3, SPT-177-Cy5 and DAPI was performed as described above and evaluated on an AxioImager.Z2 fluorescence microscope with Apotome.2.

Sample collection for microbiota profiling

For beewolf gut microbial community profiling, the following samples were collected for DNA extraction and amplicon sequencing: six beewolf larvae while feeding on the provisioned honeybees, i.e. before entering diapause, seven beewolf larvae during diapause, nine beewolf adults, eight honeybees representing the larval food source, nine nesting-sand samples (from the experimental cages), and two honey samples (used to feed adult beewolves). For microbial community profiling of cocoons, we collected eight male beewolf cocoons (from brood cells provisioned with two bees, respectively) per time point. Samples were obtained at 0, 1, 2, 4, 8, 16, 30, and 150 days after cocoon spinning using previously described methods (Koehler et al. 2013). Given the complexity of the system and limited accessibility to female cocoons, we chose to only use male cocoons in this experiment, as this allowed us to obtain a balanced set of cocoon samples from brood cells provisioned with the same number of bees. The cocoons were longitudinally cut to create an opening through which the larva was carefully removed using forceps.

DNA extraction

Guts of beewolf larvae were dissected under sterile conditions. Beewolf male adults were collected after emergence and fed with honey ad libitum for five days, then their guts were dissected. Bees from the same colonies used for provisioning the female beewolves were sampled and their guts dissected. All gut samples were stored at −80°C. Guts contained in 1.5 ml microcentrifuge sterile tubes were frozen in liquid nitrogen and homogenized with pestles, and DNA extraction was carried out with the DNeasy PowerSoil® DNA Isolation Kit following the manufacturer’s protocol.

The honey that was used for feeding beewolf adults and the sand used in the rearing cages was also sampled and processed for DNA extraction. Honey DNA extractions were carried out according to Prosser and Hebert (2017) and Balzan et al. (2020). Briefly, 10 g of honey samples were diluted in 40 ml of sterile water and incubated at 56°C for 30 min. The mixture was centrifuged at 5000 × g for 30 min using a 5810R Eppendorf centrifuge and the supernatant was discarded. The pellet was then resuspended in 25 ml of sterile water (56°C Water) and centrifuged at 5000 × g for 15 min. The pellet was then resuspended in 1 ml of 50% ethanol, transferred to a 2 ml microfuge tube, centrifuged at 12 000 × g for 15 min and dried at 56°C for 45 min. 100 mg of sterile 0.1 mm beads (Biospec, Bartlesville, USA) and 600 μl of lysis buffer were added to the dried pellet, and tubes were shaken in a TissueLyser (Qiagen, Hilden, Germany). Disruption was carried out in high-speed mode four times for 30 s at 30 Hz. Forty microliters of proteinase K (Qiagen) were added, and samples were incubated at 56°C for 90 min. After this step, the DNeasy PowerSoil® DNA Isolation Kit manufacturer’s instructions were followed.

Total DNA from soil samples was extracted as in Köhler et al. (2013) following a modified version of a previous established protocol (Berry et al. 2003). In brief, eight sand samples (30 g each) were homogenized in 50 ml of disrupting buffer (0.2 M NaCl, 50 mM Tris–HCl pH 8.0). Large particles were separated by centrifugation at 100 × g for 5 min at room temperature. The supernatant was transferred into new 50 ml tubes with Nicodenz®, and centrifuged at 10 000 × g for 20 min at 4°C. Cells were collected from the surface of Nicodenz® and washed three times with PBS. Total DNA was extracted with the DNeasy PowerSoil® DNA Isolation Kit following the manufacturer’s instructions. Positive (ZymoBIOMICS™ Microbial Community Standard, Zymo Research, USA) and negative controls (sterile water) were included during the extractions. The concentration and purity of DNA samples were analyzed using a Nanophotometer N60 (Implen, Munich, Germany).

Microbial community profiling

Bacterial 16S rRNA gene regions were sequenced by a commercial provider (StarSeq, Mainz, Germany) on a MiSeq platform (Illumina) using double indexing and a paired-end approach with a read length of 300 nucleotides. The primers 341F (5′-CCTACGGGNGGCWGCAG-3′) and 806R (5′-GACTACNVGGGTWTCTAATCC-3′) were used to sequence the V3–V4 regions of the bacterial 16Ss rRNA (Klindworth et al. 2013).

Raw sequencing data were demultiplexed using MiSeq Reporter, permitting one barcode mismatch to account for sequencing errors. Primers were removed with Cutadapt (version 5.0) (Martin 2011), and FIGARO (version 0.1) (Weinstein et al. 2019) was used to determine low-quality terminal regions. Subsequent processing was performed in DADA2 (version 1.34.0) (Callahan et al. 2016) to resolve amplicon sequence variants (ASVs). Sample-specific error rates were learned using DADA2’s machine learning algorithm. Paired-end reads were merged, and chimera removal was done with the “consensus” method. Taxonomy was assigned to ASVs with the SILVA trainset v138.1 reference database (Quast et al. 2012). Processed ASVs and their taxonomic assignments were imported into the “phyloseq” R package (version 1.50.0) (Mcmurdie and Holmes 2013) for community-level analysis. We excluded samples with a sequencing depth below 10% of the overall mean. This resulted in the exclusion of a single sample of a 2-day-old cocoon. ASVs that could not be classified as “bacteria” (e.g. chloroplasts, and mitochondrial sequences) were filtered out. Alpha diversity was assessed using the Shannon (quantitative diversity) and Simpson (dominance) indices. To account for differences in sequencing depth, ASV counts were transformed to relative abundances and used for Bray–Curtis (BC) beta diversity analyses and taxonomic bar plot visualization. Differences in community composition were evaluated by permutational multivariate analysis of variance (PERMANOVA) based on BC distances. Statistical comparisons of alpha diversity between sample groups were performed using non-parametric tests, i.e. Kruskal–Wallis tests followed by Dunn’s post hoc test with Bonferroni correction for multiple comparisons. For beta diversity analyses, PERMANOVA with 999 permutations was used to evaluate significant differences in community composition between sample groups. Prior to ordination, read numbers were converted to relative abundance (proportional) values.

Core microbiome analysis was conducted using the “Microbiome” R package (version 1.28.0), and core taxa were defined as ASVs that meet two criteria: prevalence, defined as being present in ≥80% of samples within a given sample group, and abundance, defined as having a relative abundance of >0.1% per individual sample.

The sequence data reported in this study have been deposited in the European Nucleotide Archive (ENA) at EMBL-EBI under accession number PRJEB105043.

Absolute quantification of bacterial abundance

Bacterial 16S rRNA gene copy numbers were quantified using qPCR on the same extracted samples that were used for amplicon sequencing. qPCR reactions were performed in 20 μl reactions using Blue S’Green qPCR mix (Biozym), 1 μl template DNA, and 0.4 μM of each primer. The bacterial 16S primers EUB338mod (5′- TCCTACGGGAGGCAGCAG-3′) and EUB518 (5′- ATTACCGCGGCTGCTGG-3′) (Fierer et al. 2005), and specific Wolbachia 16S primers Wolb_16S_qPCR_fwd (5′- TTGCTATTAGATGAGCCTATATTAG-3′) Wolb_16S_qPCR_rev (5′-GTGTGGCTGATCATCCTCT-3′) (Makepeace et al. 2006) were used.

Reference DNA templates were prepared by PCR amplification and purification of the 16S rRNA gene fragments from S. philanthi and Wolbachia, respectively. The DNA concentrations were measured using a Nanophotometer N60 (Implen) and the copy numbers were calculated using the formula:

A 10-fold serial dilution was prepared for each of the templates to generate a range from 10⁷ to 10² copies/µl.

DNA copy numbers in the samples were calculated from the quantification cycle (Cq) values using the standard curve equation: , where X is the estimated DNA copy numbers, b is the y-intercept and m is the slope of the standard curve.

Results

Localization of S. philanthi in P. triangulum larvae and cocoons

Histological sections were used to assess the distribution of symbionts and the morphology of beewolf larvae. By using two parallel section series, of which one series was hybridized with bacteria- and symbiont-specific fluorescent oligonucleotide probes and the other stained with HE, observations on the morphology of the larvae and the distribution of the symbionts were made simultaneously for each preparation.

First, we investigated the localization of S. philanthi in larvae that had acquired the symbionts from the brood cell but not yet started to spin a cocoon. When investigating the sections from anterior to posterior, distinct changes in the width of the lumen of the digestive tract were observed. The esophagus was accompanied by salivary gland ducts and showed a narrow lumen. S. philanthi was not detectable in the esophagus, nor in the salivary glands (Fig. S1a and d). Where the esophagus transitioned into the midgut, the lumen enlarged. In the front region of the midgut, S. philanthi was observed at the highest cell density of the whole intestine (Fig. 2a and b; Fig. S4b). The width of the intestinal lumen was maintained over almost the entire length of the subsequent intestine. Streptomyces cell density decreased sharply at the end of the anterior midgut and remained low in posterior regions (Fig. 2c and d; Fig. S4c). In the posterior half of the larvae, no Streptomyces were detected with microscopic methods (Figs S1h–j, l–n, and S4). The same pattern was observed in the 3D-reconstruction (Fig. 3). Likewise, quantitative analysis using qPCR yielded comparable results, with the highest symbiont cell density observed in the anterior midgut (Figs S4e–g and S5). The intestine also contained other bacteria (detectable with EUB-388-Cy3) and pollen, the presence of which extended over the entire length of the intestine, excluding the esophagus and rectum, respectively (Fig. 2; Figs S1 and S4).

Figure 2.

Localization of symbiotic S. philanthi in beewolf larval guts after uptake from the brood cell. Haematoxylin eosin staining (HE) and fluorescence in situ hybridization (FISH) on histological sections (8 µm) of P. triangulum larvae, (a–d) collected after ingestion of the symbiont-containing secretion, i.e. before cocoon spinning, and (e–h) collected 2 days after cocoon spinning. (a, c, e, and g) HE staining on cross-sections of the beewolf larva, from anterior to posterior, scale bars = 500 µm. (b, d, f, and h) FISH of adjacent sections. S. philanthi cells were specifically stained with SPT-177-Cy5 (green), general eubacteria with EUB-388-Cy3 (red), and host nuclei/DNA with DAPI (blue), scale bars = 50 µm.

Figure 3.

Three-dimensional reconstruction of a P. triangulum larva and distribution of S. philanthi along the midgut. The larva was collected before cocoon spinning. (a) Fluorescence in situ hybridization (FISH) on histological sections (8 µm), from anterior to posterior. S. philanthi cells were specifically stained with SPT-177-Cy5 (green), general eubacteria with EUB-388-Cy3 (red), and host nuclei/DNA with DAPI (blue). Scale bar = 50 µm. (b) Categories of the symbiont abundance in the FISH images: 0 = no symbiont cells detected; + = few isolated symbiont cells; ++ = intermediate number of symbiont cells, and +++ = high number of symbiont cells. (c) 3-dimensional reconstruction with color coding according to the abundance of symbiont cells. Scale bar = 5 mm.

In larvae that were fixed two days after spinning the cocoon, clear morphological differences to larvae collected before cocoon-spinning were observed. The esophagus, surrounded by salivary gland ducts, appeared longer than in larvae before spinning and was strongly constricted (Fig. S2a and b, d and e). It widened into the midgut (see Fig. S2c and f), which maintained a thinner lumen than that of larvae prior to cocoon spinning (see Fig. S2c, g–i, f, and k–m). S. philanthi could not be detected in any sections of the larva, while other bacteria (stained with EUB-388-Cy3) were still present (Figs S2 and S6).

In larvae that were fixed on the fifth day after spinning the cocoon (i.e. 1–2 days after excreting the gut contents), the change in the morphology of the digestive tract was even more pronounced than on the second day after spinning the cocoon. Here, too, the esophagus was strongly constricted (Fig. S3a and d). In the further course of the gut, there was no significant enlargement of the lumen along its entire length, whereas the volume of the fat body was markedly increased (Fig. S3b, c, and g–j). As in larvae collected 2 days after cocoon spinning, no S. philanthi cells were detectable in the entire larva, but other bacteria (stained with EUB-388-Cy3) were still present (Figs S3 and S7).

The localization of S. philanthi on cocoons of P. triangulum was assessed in histological sections taken from the anterior end (where the head of the larva is localized) of four cocoons. As described before, S. philanthi was detected by fluorescence in situ hybridization on the outer surface of the cocoons (Kaltenpoth et al. 2005), whereas no cells could be found on the inner surface (Fig. S8). No significant differences in symbiont distribution were observed between cocoons of male and female larvae or cocoons that were three days and two months old.

Microbial community dynamics across beewolf life stages

During the larval stage, beewolves consume paralyzed bees provided by their mothers, whereas adult beewolves feed primarily on nectar. We assessed if beewolf microbial communities at the ASV level vary throughout different developmental stages, and if so, whether this variation was influenced by their diet. We also tested whether the microbes from nesting sand (from an in vitro rearing setup) could be a source of recruitment of gut microbial community members.

We sequenced 22 samples across different life stages of beewolves (i.e. guts of six feeding larvae, seven larvae during diapause, and nine adults), as well as 10 samples of nutritional resources (eight bees and two honey samples), and nine nest samples (sand). Sample sequencing depth ranged from 4374 to 100 981 reads with an average of 25 860 per sample (total number of reads was 748 055). After filtering, denoising, and chimera removal, 284 655 reads were retained across all samples.

Feeding beewolf larvae and honeybees shared a core gut microbiome composed of seven ASVs (among them Lactobacillus apis, Bifidobacterium asteroides, and Gilliamella apicola), with >0.1% abundance and at least 80% prevalence among all samples from the corresponding group (Fig. S10). Bacilli (52.7%) constituted the most abundant class, followed by Alphaproteobacteria, Gammaproteopacteria, and Actinomycetes (18.2%, 17.5%, and 11.52%, respectively) in actively feeding larvae. (Table S). Wolbachia only made up an average of 2.6% of the bacterial community in the guts.

In stark contrast, the gut microbial communities of beewolves during larval diapause and in the adult stage were dominated by Alphaproteobacteria (Fig. 4), with the most abundant ASV classified as Wolbachia (87.9% and 81.2% relative abundance in guts of diapausing larvae and adults, respectively). This ASV was identical to the reference 16S gene from Wolbachia previously sequenced from P. triangulum (GenBank accession number NZ_OZ034757). An additional 15 Wolbachia ASVs, mostly with abundances below 0.1%, consisted of sequences with up to three mismatches to the dominant ASV. Overall, this suggests that nutritionally acquired microbes shape the microbial communities of actively feeding beewolf larvae, but this bacterial community is transient, with excretion of the gut content resulting in a microbiome dominated by Wolbachia before entering diapause.

Figure 4.

Relative abundance of the bacterial ASVs in the gut across beewolf developmental stages and in their nutritional resources and nest environment. Sample sizes are as follows: beewolf adults (n = 9), diapausing larvae (n = 7), feeding larvae (n = 6), honeybees (n = 8), honey (n = 2), and soil (n = 8). Relative abundances are indicated at the class and genus level.

Concordantly, there were significant differences in alpha-diversity (Kruskal–Wallis test: Shannon X²= 24.2 P < .0001; Simpson X²= 26.8 P < .0001) and beta-diversity (PERMANOVA: F = 6.486 P = .001) among beewolf life stages. Adults and diapausing larvae showed significantly lower alpha-diversity than feeding larvae and honeybees (Fig. S9). Correspondingly, bacterial community composition differed between life stages, with feeding larvae being similar to their honeybee food, whereas adults and diapausing larvae were similar to each other, but distinct from feeding larvae and honeybees (Fig. 5c). Differences in dispersion were not detected (BETDISPER: F = 2.049, P = .1316) when beewolf developmental stages and honeybee bacterial communities (Bray–Curtis distances) were compared, indicating that variation within groups is not responsible for the observed differences.

Figure 5.

Absolute bacterial abundance and ordination principal coordinates analysis (PCoA) on Bray–Curtis distances for the bacterial communities associated with the gut of different European beewolf life stages and their honeybee prey. (a) General eubacterial 16S rRNA gene copy numbers, and (b) Wolbachia 16S rRNA gene copy numbers. Letters indicate significant differences among groups (least-squares means pairwise comparison with Tukey adjustment, P < .05). The dashed red line represents the negative controls included in the qPCR reactions. Data were log-transformed to facilitate their visualization. (c) Beta diversity of bacterial communities of beewolf and honeybee guts, combined with soil and honey.

As we only included two samples of honey, these samples were excluded from comparisons of statistical analysis, but descriptive analyses were conducted. The most abundant ASVs in honey samples were Melissococcus plutonius, Paenibacillus larvae, Bacillus sp., and Tyzzerella sp. accounting for 78% of the taxon composition in this sample group. Moreover, 28 ASVs were found in honey samples from which 11 were also found in other groups (e.g Lactobacillus apis and Bifidobacterium asteroides in beewolf larvae and honeybees at relative abundances between 3.2% and 6.7%). Lactobacillus apis was 100% prevalent in both beewolf feeding larvae and honeybee samples and have similar relative abundances of around 5% (Fig. 4). While comparing core microbiomes with thresholds of 0.1% abundance and 50% prevalence ASV among sample groups, we could not find a single ASV from soil shared with other sample groups.

The beewolf life stages were also significantly different in absolute bacterial abundance (LM multiple, R²= 0.87, F = 65.52, P < .0001). Bacterial abundance was higher in feeding larvae compare to adults and larvae in diapause (Tukey HSD, P > .001), whereas no significant differences were observed between adults and larvae in diapause (Tukey HSD, P = 1.0) (Fig. 5a). We also assessed the abundance of Wolbachia and found it to be stable across beewolf life stages. Wolbachia abundance was significantly higher in diapausing larvae and in adults (Dunn’s test adj. P-value < .01) as well as in feeding larvae (Dunn’s test adj. P-value < .05), compared to honeybees (Kruskal–Wallis χ2 = 19.076, df = 3, P-value = 2.6e-04) (Fig. 5b).

Microbial community dynamics on beewolf cocoons

To characterize the composition of beewolf cocoon bacterial communities during the long period of larval hibernation, we sequenced 64 cocoons at eight different time points (days 0, 1, 2, 4, 8, 16, 30, and 150 after cocoon spinning). Sample sequencing depth ranged from 2192 to 302 702 reads (mean: 86 024 reads per sample; total number of reads: 5 695 387). After filtering, denoising, and chimera removal, 3 227 555 reads were left and 63 samples retained for further analysis.

As expected, the defensive symbiont S. philanthi was present across all time points and dominated the bacterial community on the cocoon in the first 1–2 weeks after cocoon spinning. It exhibited a mean relative abundance of 30% across all different time points measured in this study. Before larval gut content excretion, S. philanthi abundance ranged from 40% to 53% and then gradually decreased from day 8 until day 150 from 18% to 2.7%, respectively. This decline was accompanied by an increase in the relative proportions of Pseudomonadota (formerly Proteobacteria) (Fig. 6).

Figure 6.

Relative abundances of bacterial ASVs on beewolf cocoons from the time of cocoon spinning (day 0) until day 150 after cocoon spinning. Eight cocoons were collected per time point, but for Day 2 one sample was excluded due to low sequencing depth. Relative abundances are indicated at the class and genus level.

The “core microbiome” of beewolf cocoons across time points (days 0–150) with thresholds of 0.1% abundance and 80% prevalence was composed of only Streptomyces. With more relaxed thresholds of 0.1% abundance and 50% prevalence, the following bacterial taxa were part of the core microbiome: unclassified Rhizobiaceae, Pseudomonas, Cupriavidus, Bacteroides, Pseudarthrobacter, Wolbachia, and Paraburkholderia. Since beewolf larvae release their gut content onto the cocoon about 4 days after cocoon spinning, the microbiota changed around this time and showed an increasing abundance of bee-associated taxa. Before day 4, the major community members of cocoons (0.1% abundance and 50% prevalence) in order of abundance are Streptomyces, Bacteroides, Wolbachia, Pseudomonas, Ensifer, Ralstonia, and Cupriavidus, while after day 4 this composition shifts to Streptomyces, Lactobacillus, Rhodococcus, Bifidobacterium, Gilliamella, Ensifer, and Bartonella.

When comparing taxonomic richness in beewolf cocoons before and after the larvae emptied their gut contents (day 4 after cocoon spinning) we did not find significant differences (Kruskal–Wallis test: Shannon X² = 9.32, df = 7, P-value = .231; Simpson X² = 11.49, df = 7, P-value = .119; Fig. S11). However, significant differences in microbial community composition were observed when comparing all time points (PERMANOVA, F = 2.4693, P = .005). A pairwise permutation test results with Benjamini–Hochberg FDR-adjusted P-values showed several significant differences in community composition (FDR < 0.05), especially between very early and very late time points (Fig. S12 and Table S1).

Discussion

We investigated the transmission route of the European beewolf’s defensive symbiont from the adult female to the larval cocoon and characterized the bacterial communities associated with different beewolf life stages. Our results demonstrate that beewolf larvae acquire the symbiotic bacteria from the brood cell into the lumen of the proximal midgut and regurgitate them onto the cocoon during cocoon-spinning. The symbionts are only transiently maintained in the gut, as by day two after cocoon-spinning we were not able to detect S. philanthi cells in the beewolf larval gut anymore (Fig. 2). Thus, our findings support the hypothesis that newly eclosed females must acquire the symbionts directly from the cocoon in the time window after shedding the pupal skin and before emergence from the cocoon (Kaltenpoth et al. 2009).

The beewolf symbiont’s complex vertical transmission route including the secretion from the female’s AGRs into the brood cell, the acquisition into the larval midgut and regurgitation onto the cocoon, and the symbiont uptake by eclosing new adults relies on the ability of the symbiont cells to persist on the cocoon during prolonged periods of time. Previous studies provided evidence that the symbionts enter a stage of low metabolic activity after producing the protective secondary metabolites during the first 2 weeks after cocoon-spinning (Koehler et al. 2013). However, in contrast to free-living Streptomyces species (Mccormick and Flärdh 2012), S. philanthi does not seem to be able to produce spores in vitro (Nechitaylo et al. 2014), and genomic analyses revealed that important genes for sporulation have been pseudogenized (Nechitaylo et al. 2021). Thus, it currently remains unknown how the symbiont cells can preserve viability while facing adverse ecological conditions during the long period of beewolf hibernation. If nutrient provisioning by the host to the symbionts on the cocoon occurs, then it can happen either during transmission of the symbionts with the midgut content, via the cocoon silk, or through the release of the gut content 3–4 days after cocoon-spinning.

Beyond environmental stressors, the beewolf symbionts may face competition from the brood cell microbiota as well as the beewolf’s gut microbial community. In order to gain insights into community-level dynamics in adults, larvae, and on the cocoon, we characterized the bacterial communities by high-throughput amplicon sequencing. We found that the gut microbial communities of feeding larvae contain bacteria characteristic for the gut microbiota of their food, honeybee workers, i.e. Lactobacillus, Commensalibacter, Gilliamella, Snodgrassella, Bifidobacterium, and Frischella (Kwong and Moran 2016, Ellegaard and Engel 2019). These taxa are likely taken up while feeding on the honeybee prey and are only transiently present in the beewolf gut, as they are lacking in larvae hibernating in the cocoon. While the honeybee gut microbiota makes important contributions to the bees’ fitness by aiding the digestion of pollen polysaccharides, detoxifying plant secondary metabolites, and providing defense against intestinal parasites and pathogens (Engel et al. 2012, Kwong and Moran 2016), at least the dietary contributions are unlikely to be relevant for the beewolf larvae, given their carnivorous lifestyle.

The beewolf larval gut microbiota undergoes a drastic compositional rearrangement from feeding to hibernating stage, shifting to a simple community dominated by Wolbachia in hibernating larvae and adult stages. Like other apoid wasps and solitary bees, beewolves release their gut content as preparation for entering the diapause phase (Santos et al. 2019). This is likely the key driver of the observed changes in the microbial community in beewolf wasps preparing for diapause. This process diminishes or clears out non-beneficial microbes and purges microbial growth substrates and toxic metabolites, reducing the risk of pathogen infections and oxidative stress while entering diapause (Dittmer and Brucker 2021, Izadi 2025). Across different insect taxa, it has been shown that microbial communities change before and during diapause (Liu et al. 2016, Wang et al. 2019, Didion et al. 2021). These alterations in microbial profiles are characterized by a reduction in microbial diversity and in some cases selection of a diapause-specific microbiota (Arias-Cordero et al. 2012). Our results show that after gut content release, only Wolbachia persists in the beewolf gut. This persistence may be due to Wolbachia’s localization in the gut epithelial cells rather than in the lumen, which is consistent with our observations that Wolbachia abundance does not change between feeding and diapausing larvae. Wolbachia is the most prevalent intracellular reproductive manipulator in insects, and it can also contribute to host protection against RNA virus infections, like in the case of mosquitoes and Zika and Dengue viruses (Rajendran et al. 2024). Additionally, nutrient provisioning by Wolbachia has been reported under stress conditions in D. melanogaster (Brownlie et al. 2009) as well in grasshoppers (Ju et al. 2020). However, whether Wolbachia impacts beewolf fitness by reproductive manipulation, defense, or via the supplementation of limiting nutrients (Kaur et al. 2021) currently remains unknown.

On the beewolf cocoon, the bacterial community is dominated by S. philanthi, but experiences a shift in relative abundance from Actinomycetota to Pseudomonadota during beewolf hibernation. Previous studies have shown that the S. philanthi population size increases during the first two days after cocoon spinning and then remains stable, but after 1–2 weeks the symbionts enter a dormant phase characterized by the reduction of gene expression including housekeeping genes (Koehler et al. 2013). Taking together, these results suggest that the gut content excretion four days after cocoon spinning represents a major disturbance event that reshapes the microbial community of beewolf cocoons, establishing a sustained compositional shift during host diapause. Intriguingly, the amount of antibiotics produced by the beewolf symbionts sharply increases 4–8 days after cocoon spinning (Koehler et al. 2013), raising the possibility that the release of the gut content and the gut microbiota onto the cocoon triggers antibiotic production by the symbionts.

By the end of the hibernation period, we could also detect other Streptomyces ASVs not corresponding to the beewolf symbiont. Thus, there may be competition between the symbionts and free-living Actinomycetes taxa for access to the beewolf-associated environment (see Kaltenpoth et al. 2014). Beewolf symbionts are known to produce a cocktail of antibiotics on the cocoon composed primarily of piericidin derivatives and streptochlorin (Kroiss et al. 2010). Previous bioassays have shown that these metabolites inhibit the growth of pathogenic fungi as well as Bacillus subtilis and Paenibacillus larvae (Kroiss et al. 2010). Moreover S. philanthi also contains additional cryptic biosynthetic gene clusters (BGCs) in its genome (Nechitaylo et al. 2021). These cryptic BGCs may be activated under specific conditions including the presence of specific pathogens and competitors. However, it is also possible that microbes beyond S. philanthi contribute to the protection of the beewolf larva in the cocoon against pathogenic fungi.

Our study expands on the characterization of the symbiont transmission route and the microbiota dynamics across different developmental stages in the European Beewolf, a charismatic example of a defensive symbiosis. Beyond yielding a glimpse into the challenging life of a beewolf symbiont—being swallowed by the larva, regurgitated onto the cocoon, and defecated upon four days later—it also provides evidence for the necessity of special adaptations to survive the nutrient-limited conditions on the cocoon in the face of microbial competitors. A better understanding how externally transmitted symbionts in beewolves and other insects navigate the challenging transitions between sheltered host organs and the harsh environment will undoubtedly yield fascinating insights into microbial adaptations to biphasic life cycles.

Conflicts of interest

The authors declare no conflict of interest.

Funding

The authors gratefully acknowledge financial support from the Deutsche Forschungsgemeinschaft (DFG; German Research Foundation) to M.K. (DFG, KA2846/7–1 as part of the Research Unit FOR 5026 “InsectInfect”), and the Max Planck Society.