Consequences of Nosema apis infection for male honey bees and their fertility

The queens of eusocial bees, ants and wasps mate only during a very short period early in life and males therefore produce ejaculates consisting of large numbers of high quality sperm. Such extreme selection for high fecundity resulted in males investing minimally into their somatic survival, including their immune system. However, if susceptible males are unable to protect their reproductive tissue from infections, they compromise queen fitness if they transfer pathogens during mating. We used the honey bee Apis mellifera and investigated the course of infection of the sexually transmitted pathogen Nosema apis. We predicted that honey bee males are susceptible but protect their reproductive tissues from infections. We investigated the effects of N. apis infections on the midgut, the accessory glands and the accessory testes and quantified the consequences of infection on male survival and fecundity. We found that N. apis is able to infect males, and as infections progressed, it significantly impacted fertility and survival in older males. Even though we confirm males to be able to minimize N. apis infections of their reproductive tissues, the parasite is present in ejaculates of older males. Consequently N. apis evolved alternative routes to successfully infect ejaculates and get sexually transmitted.

The mating biology of eusocial ants, bees and wasps is truly spectacular because reproductive females (queens) only copulate during a very brief period early in their life when they acquire and store a lifetime's supply of sperm [1][2][3] . As queens never replenish sperm later in life, sperm use and economy has been taken to inimitable extremes in those species that maintain colonies with millions of workers [3][4][5][6] and/or survive in the field for decades 4,7,8 . To provide queens with sufficient numbers of sperm, social insect males evolved ejaculates that contain large numbers of sperm of exceptionally high quality 3,9 . The production and maintenance of such ejaculates prior to mating induces substantial costs to eusocial insect males 5,10 . They are therefore thought to operate at their physiological limits, where even the slightest disturbance or alteration in their environment can compromise their fertility 11 . Additionally, to maximise their reproductive potential, these males have been found to only minimally invest into their somatic survival 5,10 .
The immunity of social insect males is also affected by their genetics because they originate from non-fertilized eggs and are haploid animals, which has been hypothesized to additionally increase their susceptibility to parasites 12 . Therefore, it is particularly interesting to study immune challenges resulting from parasitic infections in eusocial insect males. Empirical work has confirmed that immune responses of social insect males are consistently lower than those of female workers e.g. in leaf cutter ants 13 , wood ants 14 or bumblebees 15 . The resulting susceptibility of social insect males to parasitism is thought to have resulted in a number of life history adaptations, which reduce a male's risk of getting infected: e.g. they have short life spans, are completely dependent on workers and do not provide any help for their colony. However, social insects are known to host a range of parasites, which easily spread through their colonies consisting of large numbers of related individuals including males 16,17 . Male susceptibility might therefore reduce a colony's opportunity to control infections and compromise colony fitness further, especially if such infections reduce male fecundity, life expectancy or competitiveness. Finally, male susceptibility could also allow pathogens to spread through their bodies and eventually infect their reproductive organs. In such a case, sexually transmitted pathogens could then inflict cross-generational costs. However, sexually transmitted diseases and their consequences have not been studied in great detail in social insects [18][19][20] . Here we hypothesize that social insect males should minimize the risk of spreading a disease to queens by protecting their reproductive tissues from parasite infections.
To test this idea, we used the honey bee A. mellifera ligustica and investigated the effects caused by an infection with the widespread fungal disease Nosema apis. Honey bees make interesting study organisms to unravel effects of infections on male immunity and fecundity for a number of reasons. First they are known to host a wide variety of parasites 21,22 and second many of these are widespread and abundant enough among colony members to pose a potential threat to males (drones). Additionally, deformed wing virus was successfully identified in honey bee ejaculates, suggesting that some honey bee diseases could be sexually transmitted 19,23 . More recently, N. apis has been reported to be present in ejaculates of honey bee drones, and artificial insemination experiments confirmed that this pathogen can in principle be transmitted during mating 24 . Consequently, N. apis is an ideal study organism to test the effects of parasites on male fertility, and whether male susceptibility allows a disease to spread to both the reproductive organs and the ejaculate. In order to test this, we conducted a series of experiments to unravel the effects of a N. apis infection on a male's somatic tissue, by inspecting infected somatic tissues of drones at different ages and comparing the survival of infected and uninfected drones. We also investigated whether males are able to protect their reproductive organs and their ejaculates from N. apis infections to minimize the risk of transmitting infective spores during copulation. To do this we measured the effects of a N. apis infection on the drones' reproductive organs at different ages, and quantified the consequences of such an infection on their fecundity.

Materials and methods
Nosema apis spore collection. To collect N. apis spores, we sampled 20 foraging workers from five different N. apis infected honey bee colonies housed at the University of Western Australia and froze all the 100 bees we sampled for 2hrs at -20 °C. We dissected out their midguts, pooled them in an Eppendorf tube containing 1 mL of DDI water and a 3 mm tungsten bead (Qiagen, Australia), and homogenized them in a mixer mill (Retsch MM301, Australia) for 30 s at 25 Hz. Next, we layered 0.5 mL of the homogenate onto 1.5 mL of 100% Percoll (Sigma-Aldrich, Australia) in a 2 mL Eppendorf tube, and centrifuged it at 18,000 x g for 60 min at 4 °C, discarding the supernatant. The pellet containing the N. apis spores was resuspended using 1.5 mL DDI water. The sample was briefly vortexed and centrifuged again at 20,700 x g for 5 min. We repeated this procedure three times before resuspending the final pellet in 0.5 mL of DDI water to store at -80 °C. Prior to infecting drones, we thawed the sample and diluted it with sugar syrup to a final concentration of 10,000 spores/μ L. Our previous work showed that collecting N. apis spores as described above minimizes the effect on their viability 25 . Drone breeding and infection. In summer 2010, we bred drones in two different colonies according to standard apicultural practices by providing each colony with one empty drone comb. Two days prior to hatching, we placed the drone frames into an incubator at 33 °C and 90% humidity and collected drones on the day they emerged (day 0). The first 21 drones collected were used for the following procedures: The faeces of three newly hatched drones was inspected under a microscope for the presence of N. apis spores. Next, we freeze-killed 6 newly hatched drones for 2 hours to dissect out their midguts and microscopically check the gut tissue for evidence of a N. apis infection. During the dissections we found that their midguts were translucent and detected no signs of a N. apis infection, such as spores or morphological changes. In addition, we freeze-killed 6 newly hatched drones to prepare histological samples by embedding their sexual organs (accessory glands and testes) and inspect them for signs of an infection with N. apis. Finally, we froze a further 6 newly hatched drones to determine the presence or absence of N. apis DNA using microsatellite markers. As we never detected any signs of N. apis in any of the 21 newly hatched males inspected (Table 1), we concluded that newly hatched males were not infected with N. apis, which is in accordance with the literature 26 .
All remaining eclosing drones were collected and each was fed with 1 μ L (10,000 spores) of N. apis spore solution (prepared as described above), before we returned them to their maternal colonies. We recaptured 21 drones on each day 6,9,13,15,20, and 25 days post infection, and used them for the same experiments as described above: 3 drones were immediately checked for spores in their faeces, 6 drones were freeze-killed to dissect and inspect their midguts, 6 drones were used for histological embedding and inspection of their sexual organs, and 6 drones were checked for N. apis DNA using microsatellite markers. In addition, to investigate, whether sexually mature drones would be able to transmit N. apis spores to the queen during mating, we microscopically inspected ejaculates of 3 infected drones on days 13, 15, 20 and 25 days post infection, respectively. An overview over the sample sizes used for each experiment is provided in Table 1. Histological confirmation of N. apis spore presence in reproductive tissue. After collecting and freeze-killing 6 drones for each of the 7 age groups as described above, we dissected their reproductive organs (accessory glands and accessory testes). To avoid contamination, we cleaned all equipment after each dissection with ethanol, bleach and water 27 . Furthermore, we rinsed each tissue sample three times prior to histological embedding in Hayes solution (0.15 M NaCl, 1.80 mM CaCl 2 , 2.68 mM KCl, 1.19 mM NaHCO 3 , adjusted to pH 8.7 using NaOH, filtered (0.22 μ m Millex® GP filter unit, America). Next, we transferred each sample into a 2 mL tube and covered it with 1.8 mL cold fixative (2.5% glutaraldehyde, 2% paraformaldehyde) before placing it on a Stovall Belly Dancer TM Shaker on 10 rpm for 30 min on ice at 4 °C overnight in a dark box. We then transferred the samples into 0.1 M phosphate buffer at pH 7.4, mixing them on ice for 15 min on a Belly Dancer on 10 rpm. Next we placed them into 1% osmium tetroxide (Sigma-Alderich, Australia) for 2 h before rinsing them in phosphate buffer. To dehydrate the tissue samples, we first placed them in an ascending series of ethanol concentrations (50, 70, 90, and 100%) followed by two washes in propylene oxide and embedding into Epon Procure 812-Araldite resin (Polyscience, Inc., Australia), using four steps of increasing concentrations of propylene oxide to resin ratios (1:3, 1:1: 3:1 and pure resin), all according to the manufacturer's protocol. We finally placed each tissue sample in moulds containing resin at 60 °C overnight. All embedded organs were sectioned with glass knifes using an ultramicrotome (LKB). The sections of 1-2 μ m were then de-plasticized using an aged, saturated solution of sodium hydroxide in 100% ethanol (sodium ethoxide) for 5 min. Afterwards, we rinsed them in 100% ethanol and water for 10 s each before placing them in 1% w/v hydrogen peroxide for 7.5 min. Finally, we rinsed each section under tap water for 3 s and stained them with Slidder's Hematoxylin, Eosin and 1% w/v Biebrich scarlet 28 . Last we mounted the sections on microscopic slides using water-free mounting media (Entellan®, Merkck Millipore, Australia) and investigated them under an Olympus BX51 microscope (Olympus, Japan bright field, Namarski optics with UPLAN FL objective lenses), taking digital photos with an Olympus DP70 camera.
Detection of N. apis in male reproductive tissue using microsatellites. We used specific primers to test for the presence of N. apis in the drones' reproductive organs. To do this we dissected the accessory glands and accessory testes from 6 drones per age group (see Table 1) and rinsed them in Hayes solution. To extract DNA, we placed each sample in 95% ethanol before transferring it into a vial with 500 μ L extraction buffer (0.1 M Tris, 0.05 M EDTA, 0.5 M NaCl, 1% w/v Polyvinyl-pyrrolidone) and a 3 mm tungsten bead (Qiagen, Australia). After homogenising each sample for 3 min in a mixer mill at 25 Hz, we added 66 μ L of 10% sodium dodecyl sulphate before centrifugation at 20,800 x g at 4 °C for 15 min. We added 445 μ L of isopropanol to 600 μ L of the supernatant and incubated the samples on ice for 15 min before centrifugation at 20,800 x g at 4 °C for 15 min. We resuspended the DNA pellet in 500 μ L of 70% ethanol and centrifuged each sample again at 20,800 x g at 4 °C for 15 min. After discarding the supernatant, each pellet was air dried for 20-45 min before being resuspended in 100 μ L of DDI water and centrifuged again at 20,800 x g at 4 °C for 15 min. We collected the supernatant and determined the amount of DNA using a NanoDrop (ND-1000 V3.2.1., America). Prior to PCR, we diluted each sample to a final concentration of 50 ng DNA/ μ L in DDI water.
The effect of N.apis infection on drone survival and sperm viability. In 2013, we conducted an additional experiment to quantify the effect of N. apis on drone survival and sperm viability at different ages, and bred drones in two different colonies as described above. After hatching, we randomly allocated 20-30 drones to one of 25 cages (14 × 19.5 × 2.3 cm, punctured metal sheet on one side and drone excluder on the other) and placed them back into their maternal colonies for the next 24 h. We recollected the cages and fed the drones with either 1 μ L Nosema spore solution (10,000 spores), or with 1 μ L of 100% w/v sugar syrup as a control. Consequently we ended up with 12 cages containing infected drones and 14 cages containing uninfected drones. We resampled drones 12, 13, 16, 19, 20, 24 and 25 days post infection. For each cage, we first counted all live and dead drones to quantify drone survival. We anesthetized the surviving drones in chloroform to initiate ejaculation and gently squeezed the drones' abdomen between two fingers, until semen appeared at the tip of the endophallus 30 . To quantify sperm viability, we used flow cytometry as described by Paynter et al. (2014) 31 . In short, we collected around 2 μ L of ejaculate per drone in 1 μ L of semen diluent (188.3 mM sodium chloride, 5.6 mM glucose, 574.1 nM arginine, 684.0 nM lysine, 50 mM tris (hydroxymethyl) aminomethane, pH 8.7). We added 1 mL of semen diluent to each semen sample and gently mixed it by turning the vial upside down until the ejaculate had fully dispersed. To avoid mucus clogging up the capillary of the flow-cytometer, we filtered a subsample of 200 μ L through a 50 μ m diameter nylon mesh and added 800 μ L of semen diluent to the filtered sample. To differentially dye live and dead sperm, we used 400 μ L of sperm sample, added 2 μ L of 1 mM SYBR 14 dye (Invitrogen, cat no. L-7011) and incubated it in the dark for 10 min. We added 2 μ L of 2.4 mM Propidium Iodide (PI) (Invitrogen, cat no. L-7011) and incubated the sample for 7 min in the dark. Sperm viability was quantified for a minimum of 3000 sperm in a BD FACS Canto II digital flow cytometer (America). The flow cytometer recorded SYBR 14 fluorescent emission in the range 515-545 nm, and PI fluorescent emission in the range 670-735 nm without compensation for spectral overlap. We recorded height rather than area of the voltage pulse generated by SYBR 14 and PI, because honey bee sperm is exceptionally long (260 μ m) 1 . Sperm dyed with SYBR 14 were gated as live and those stained with PI were gated as dead, using the FlowJo software package Version 7.6.5 for Windows (TreeStar, USA). We used the autogate function of the Flowjo software, except in 10 cases where the populations of live and dead sperm could not be separated by the software. As is typically done in these cases we therefore assigned the gates manually. We ignored doubly stained cells, which occurred at very low frequencies. Statistical analyses were performed using SPSS version 21 for Macintosh. Drone ages were grouped for Figures, but statistical analyses were done using drone age as a covariate.

Results
We found that all drones fed with N. apis spores consequently developed an infection (Table 1), which we were able to confirm by the presence of morphological changes that became visible in the midgut of infected compared to non-infected drones (Fig. 1). We found that the midguts of newly hatched, non-infected drones appeared transparent and remained translucent as drones matured (Fig. 1A,C). In contrast, as the infected drones matured, their midguts developed substantial swelling, lost their transparency and changed their colour to a grey-white (Fig. 1B,D). Microscopic inspections confirmed that the epithelial cells of the midguts of infected drones were filled with N. apis spores ( Fig. 2A-C) and we could observe large numbers of newly-released spores from burst cells (Fig. 2D).
When we compared dissected accessory glands and accessory testes of infected and non-infected drones (Fig. 3) we did not find any of the morphological signs of infection as described above for the midgut tissue ( Fig. 1 & 2). Furthermore, we never found any N. apis spores to be present in the sexual organs of infected drones of any age, neither within the tissue nor in the lumen containing either sperm or seminal fluid. This observation was confirmed histologically when comparing the sexual tracts of non-infected and infected drones of different ages. We found that the muscular tissue surrounding the accessory gland and its epithelial cell layer gradually degenerated with increasing drone age (Fig. 3), which was not the case for the muscular tissue surrounding the accessory testes (Fig. 3). However, neither the accessory glands nor the accessory testes tissues showed any visible signs of infection (Fig. 3). Furthermore, despite a very careful microscopic inspection of both the accessory testes and accessory glands, we never found a single N. apis spore in any of the infected males. However, our PCR analysis detected N. apis DNA in male reproductive tissues (Fig. 4). This was the case for all drone age groups except for the newly emerged drones, confirming that newly eclosed drones are not infected with N. apis.    Finally, when we microscopically inspected ejaculated semen, N. apis spores were only present in drones aged 20 and 25 days (Fig. 5), but never in younger drones at 9, 13 or 15 days of age.

Effect of N. apis infection on drone fecundity and survival.
Sperm viability data became available for 49 infected and 60 non-infected males. We found that drone survival significantly decreased with increasing age (ANCOVA, P < 0.001, see Table 2 for statistical details, Fig. 6 A).There was a significant effect of N. apis infection on male mortality, indicated by a significant treatment x age interaction (ANCOVA, df = 1, P = 0.013, Fig. 6 A). Drone survival was similar between infected and non-infected drones until they reached an age of about 16 days, after which mortality in infected drones substantially increased compared to the control treatment. Pairwise post-hoc t-tests using age groups revealed that a significantly lower proportion of treated drones (20.7%) per cage survived to the age of 24 to 25 days compared to control drones (54.7%), with t = 3.5, df = 6 and p = 0.013. The results of the post-hoc tests for the other age groups were non-significant.
Similarly, sperm viability also decreased with male age (ANCOVA, P = 0.003, see Table 3 for statistical details, Fig. 6 B). We also found a significant treatment x age interaction term, indicating that infected males lost their fertility faster than control drones (ANCOVA, P = 0.013, see Table 3). The results of individual, pairwise post-hoc t-tests using age groups were not significant.

Discussion
Our experiments revealed that N. apis is able to infect drones, and that these infections built up to a point where they induced significant costs for males. We found a reduction in fertility and life span as   Table 2. Survival of uninfected honey bee males and males infected with Nosema apis. Results from an analysis of covariance examining the effects of an infection with the microsporidian parasite Nosema apis on the survival of male honey bees at different ages. Dependent variable: % of drone survival per cage. In a first analysis we used treatment as fixed factor, colony as random factor and age as a covariate. Because colony was non significant (F = 0.210, df = 1, P = 0.727), we consequently removed colony as a factor for the analysis shown. a. R Squared = .799 (Adjusted R Squared = .771). drones aged (Fig. 6), as well as ejaculates of infected drones becoming contaminated with spores (Fig. 4). Honey bee drones become sexually mature from 12 days after emergence and consequently maintain their maximal fertility potential over a time period of approximately 10 days 32 . During that time, they participate in nuptial flights in order to find and mate with virgin queens. The phenotypic expression of a N. apis infection such as spores in the ejaculate and a reduction of drone fertility and survival therefore affects these drones during their main reproductive period. These findings indicate that drones which become infected shortly after hatching will eventually face substantial fitness costs and pose an infection threat to virgin queens in case they mate. It would therefore be interesting to investigate, whether infected drones indeed leave their colonies for nuptial flights and if so, whether their mating success is compromised compared to that of non-infected males. From a colony's perspective, infected males might choose not to participate in mating flights and mating in order not to compromise the mating success of their non-infected brothers. Such "altruistic" self-removal has been reported for honey bee workers after prolonged CO 2 narcosis or fed with the cytostatic drug hydroxyurea. Both treatments increased worker-mortality, and surviving foragers left their colonies, effectively committing altruistic suicide 33 . Such self-removal is also known from other social insects 34 . More research is required to quantify the risk of vertical transmission posed by males infected with N. apis. Although our visual inspections, both morphologically as well as histologically, did not reveal any obvious signs of N. apis infections in reproductive tissues or their products, DNA of N. apis can be detected in both accessory glands and accessory testes. These findings indicate that N. apis is able to establish low levels of infections in reproductive tissues of drones. However, as we did not find spores in any of the reproductive tissues we inspected, the pathogen seems unable to complete its reproductive cycle or to build up an infection. Therefore, even though N. apis is able to infect drone accessory testes and accessory glands, the drones seem able to slow down or prevent this parasite from producing spores within their reproductive tissue, thereby reducing the risk of sexual transmission. This is an interesting finding, because N. apis infections are already known to spread to different honey bee organs such as the fat body, the malpigian tubules or the haemolymph 35 . In comparison, Nosema ceranae has been reported to infect drones at the pupal stage already 36 , to reduce drone body weight and life span and to induce physiological changes in honey bee queens 37 . Even though drones are more susceptible to N. ceranae than workers, they were are to develop higher tolerance to N. ceranae in specifically selected honey bee strains 38 . It would therefore be interesting to study how males are able to protect their reproductive tissue, and whether the costs associated with suppressing N. apis infections result in the observed decrease in sperm viability and male survival.
We did find N. apis spores in the semen of older drones, which raises the question of how they were able to contaminate the ejaculate. Our dissection work revealed that N. apis inflicts substantial damage to tissues such as the midgut, which becomes increasingly more fragile with age and can easily be damaged. Furthermore, N. apis is known to cause dysentery in honey bees, resulting in spore-containing faecal residues on bees 39,40 . It is therefore possible that the spores we detected in the ejaculate of older drones are either caused by faecal contaminations of the endophallus or haemorrhaging after tissue damage. Ejaculation is a traumatic process as the drone contracts its abdominal muscles in order to build up haemolymph-pressure, inducing the irreversible expulsion of the endophallus and then the ejaculate. During a final step of ejaculation, the tip of the drones' endophallus bursts, breaks off and is left inside the queen after copulation 41 , causing the drones' death. It therefore seems reasonable to assume that infected tissues, which are more prone to damage and tear already, release N. apis spores under the pressure build-up described above, resulting in spore contamination of the ejaculate. Further experimental work is required to unravel these proximate mechanisms. However, if infections occur during the ejaculation process, a male's only counter measure to protect his ejaculate and mate would need to be derived from immune traits present within the ejaculate. Interestingly, drone ejaculates not only consist of sperm, but also of substantial amounts of seminal fluid, produced by the accessory glands 1,42 . The latter is biochemically complex and contains a number of immune proteins, some of them with well-known antifungal properties such as chitinase 43 . Further research should therefore test the idea that seminal fluid is able to kill N. apis spores, in order to minimize the risk of N. apis establishing an infection inside the queen.