Pathogen isolation

Twenty Danish A. apis strains were isolated from honeybees mummies collected by Danish bee keepers. Infected larvae were surface sterilized in 10% sodium hypochlorite for 10 min followed by 2 min. water washing [13]. Rinsed larvae were cut into three pieces and placed on Sabouraud Dextrose Agar (SDA) growth medium at 34°C. After several days the A. apis mycelia were observed growing on the agar plates. Single hyphal tips were isolated with a sterile scalpel using a dissecting microscope. Each hyphal tip was placed on a new Petri dish with SDA growth medium, incubated at 34°C until growth was observed and stored at 25°C for 2 weeks. For long term storage mycelia were placed in 20% glycerol at −80°C [14].

DNA extraction, PCR amplification and sequencing

Genomic DNA from A. apis isolates was extracted from lyophilized hypae using the DNeasy® Plant Mini Kit (Qiagen) and DNA extracts were diluted 1∶10 in sterile MilliQ water prior to polymerase chain (PCR) reaction. PCR amplification was conducted for a variable part of the EF1α and two intergenic regions located on scaffolds 300 and 1635 of the assembled A. apis genome sequence [15]. Samples for PCR amplifications consisted of 1 U Phusion® High-Fidelity DNA Polymerase (New England Biolabs, Inc.) with appropriate buffer (HF buffer (1.5 mM MgCL2), 0.2 µm dNTPs, 1 µm of each forward and reverse primer, in a final reaction volume 50 µL. All reactions were carried out on a T1 Thermocycler using a touchdown approach with cycling conditions consisting of: 30 s denaturation at 98°C; 10 cycles at 98°C for 30 s; 70–60 cycles (decrease of 1°C per cycle) for 30 s and 72°C for 30 s; 30 cycles of 98°C for 30 s, 60°C for 30 s, 72°C for 30 s, with a final 10 min extension at 72°C. PCR products were electrophoretically separated on 1.5% agarose gels, visualised with EZ vision One® (Amresco), cleaned with an illustra GFX™ PCR DNA and Gel Band Purification Kit (GE-Healthcare) and sent to Eurofins MWG Operon AG, Ebersberg, Germany for sequencing with both forward and reverse primers.

DNA sequence analyses

Sequences were edited and aligned manually using BioEdit [16] and sequence analysis of alignments were conducted in MEGA version 4 [17] on a dataset combining all three loci (EF1a, scaffold 300 and scaffold 1634) using the Neighbor-Joining method with a pairwise deletion option. Evolutionary distances between strains were computed using the Jukes-Cantor method and branch support values were assessed by bootstrapping of 1000 replicate datasets. Further information on the strain collection can be found in Table S1.

Maintenance of Ascosphaera apis cultures and inoculum preparation

Ascosphaera apis is a heterothallic fungus, meaning that production of spores only occurs when the hyphae of both mating types are in contact. Therefore each isolated strain had to be mated with the characterized strain ARSEF 7405 or ARSEF 7406 (USDA-ARS Collection of Entomopathogenic Fungal Cultures in Ithaca, New York, USA). Once the strains were designated a mating type, they were paired and placed on a Petri dish. We chose 4 pairs of strains designated A (KVL06-150, KVL06-158), D (KVL06-182, KVL08-41), F (KVL06-123, KVL06-132), and G (KVL07-087, KVL07-104). The two paired strains A and D came from one phylogenetic clade and the paired strains F and G from another clade (see below).

In order to obtain fresh spores 3 weeks prior to the experiment, isolated strains from each phylogenetic clade were paired. The produced spores were removed from the plates with a small sterile spatula and placed into a sterile glass grinder with 20 µl of sterile deionized water. Following the grinding, 50 µl of sterile deionized water was added to the spore suspension. Large particles in the suspension were allowed to settle for 20 min, and a sample of approximately 50 µl was taken from the middle of the suspension. Spore concentration in the resulting suspension was determined with a hemocytometer (Tiefe Depth Profondeur, Marienfeld, Germany).

Spore viability

Spore viability for A. apis was tested following the protocol of James and Buckner [18] with a few modifications. A spore suspension (150 µl) of a concentration of 2×10⁷ spores per ml was mixed with 150 µl GLEN, a liquid medium suitable for germination and in vitro growth of insect pathogenic fungi [19]. Droplets of 10 µl of mixture were placed on three spots of a sterile Teflon coated slide, which was deposited in a sterile Petri dish lined with wet filter paper. Each Petri dish was subsequently placed in an airtight container flushed with CO₂. The containers were incubated for 24 hours at 34°C, after which the Teflon coated slide received a cover-slip and the spore germination percentage was determined using differential interference contrast microscopy at 400x magnification. One hundred spores were evaluated for enlargement or germ tube formation in three different randomly chosen fields of view. Overall, the spore germination rates ranged from 10 to 20%.

Host maintenance and in vitro rearing

Honeybee (A. mellifera) larvae were obtained from an apiary located at the University of Copenhagen. Colonies were checked regularly and were free of any noticeable brood and adult bee diseases. For each experiment larvae were transferred from the three hives and reared in vitro following the protocol of Aupinel et al. [20] and Vojvodic et al. [12] with a few modifications. Larval age was estimated by size [21] and larvae that were 24 h old (+/− 6 hours) were taken from the combs using a Swiss grafting tool (Swienty, Sønderborg, Denmark). After removal from the comb each larva was placed into an individual cell of a 48-well tissue culture plate with 10 µl of larval diet. The larval diet consisted of 50% of Chinese fresh frozen royal jelly (v/v) (Sonnentracht Imkerei GmbH, Bremen, Germany), 6% D-glucose (w/v), 6% D-fructose (w/v), 1% (w/v) yeast extract and sterile deionised water. The diet was mixed and frozen in smaller aliquots and was pre-heated to 34°C before being used for feeding. The larvae were fed once a day with 20 µl on the first three days, and 40 µl on day four. The tissue culture plates with the larvae were stored in a humid chamber and incubated at 34°C in constant darkness. Wells were gently cleaned with cotton wool in case larvae started to defecate.

Host inoculation

Two days before the experiment larvae were removed from each of three hives with unrelated queens and reared in vitro as described above. After a 48 h acclimatization period, 30 healthy larvae were fed 5 µl of a designated spore suspension of one of the pathogen strains (A, D, F, or G) using 5×10⁵ spores/ml and distilled water (in the case of the control). In total, 360 honeybee larvae from three hives were exposed to one of the 4 genetically distinct A. apis strains and 90 larvae were treated with distilled water as a control. To avoid any temporal and environmental differences the experiment was set up at one time period, limiting the number of colonies as well as individual bees that could have been handled simultaneously. Within one day, the larvae had ingested all food, including the spores. The possibility of spores present after day one, and the risk of later infections were thus minimized. The experimental larvae were kept in a humid chamber at a constant temperature of 34°C for 7 days. The number of diseased, surviving, and infected larvae were examined microscopically and recorded daily. Infected host larvae were identified by ceased respiration, loss of body elasticity, or a change to gray or brown colors, and fungal hyphae on the cuticle. Larvae that died without any visual presence of fungal hyphae were re-examined the following day. If the pathogen was observed protruding through the host cuticle, these larvae were considered dead from the pathogen on their initial day of death. If the pathogen was not visually present on dead larvae, they were recorded as dead from natural causes.

Statistical analysis - Survivorship analysis

Statistical analyses were done using the proportional hazard model (also known as Cox regression) analyzing the event times at the day of death, and censoring times at the termination of the study on day 8 [22]. In this model the instantaneous hazards of dying from the infection were described as functions of time, colony, and strain. The proportional hazard null model was that the instantaneous hazards were proportional across hives and strains when considered as functions of time. Prior to formal hypothesis testing, this assumption was validated using the methods proposed in Lin et al. [23]. Model reduction was done using likelihood ratio tests starting from the initial model including the main effects of hive and strain together with their interaction. Post-hoc comparisons were done using Wald tests. Since the larvae were observed only once a day, several larvae were sometimes observed to have died at the same time. Such observational ties were analyzed by averaging over the event times and under the assumption that the censored times have taken place after the event times. All computations were done using SAS V9.2. Bonferonni correction for multiple testing was used to adjust the reported p-values for post-hoc comparisons.