Host sharing by the honey bee parasites Lotmaria passim and Nosema ceranae

Abstract The trypanosome Lotmaria passim and the microsporidian Nosema ceranae are common parasites of the honey bee, Apis mellifera, intestine, but the nature of interactions between them is unknown. Here, we took advantage of naturally occurring infections and quantified infection loads of individual workers (N = 408) originating from three apiaries (four colonies per apiary) using PCR to test for interactions between these two parasites. For that purpose, we measured the frequency of single and double infections, estimated the parasite loads of single and double infections, and determined the type of correlation between both parasites in double infections. If interactions between both parasites are strong and antagonistic, single infections should be more frequent than double infections, double infections will have lower parasite loads than single infections, and double infections will present a negative correlation. Overall, a total of 88 workers were infected with N. ceranae, 53 with L. passim, and eight with both parasites. Although both parasites were found in all three apiaries, there were significant differences among apiaries in the proportions of infected bees. The data show no significant differences between the expected and observed frequencies of single‐ and double‐infected bees. While the infection loads of individual bees were significantly higher for L. passim compared to N. ceranae, there were no significant differences in infection loads between single‐ and double‐infected hosts for both parasites. These results suggest no strong interactions between the two parasites in honey bees, possibly due to spatial separation in the host. The significant positive correlation between L. passim and N. ceranae infection loads in double‐infected hosts therefore most likely results from differences among individual hosts rather than cooperation between parasites. Even if hosts are infected by multiple parasites, this does not necessarily imply that there are any significant interactions between them.


| INTRODUCTION
Host-pathogen interactions are a major driving force of evolution and have received considerable attention (Olive & Sassetti, 2016;Rau et al., 2015;Woolhouse, Webster, Domingo, Charlesworth, & Levin, 2002). Although hosts infected by more than one parasite are common (Bordes & Morand, 2011), further attention is required to address interactions among such parasites within one host. These parasite-parasite interactions in individual hosts can potentially range from competition to cooperation (Alizon & Lion, 2011;Dobson, 1985;Griffin, West, & Buckling, 2004;Poulin, 2001;Read & Taylor, 2001;West & Buckling, 2003) and may have tremendous effects. For example, a host may cope with a single infection, but succumb to multiple ones, depending on pathogen-pathogen interactions (Neumann, Yañez, Fries, & de Miranda, 2012;Shen, Yang, Cox-Foster, & Cui, 2005). This creates demand for better understanding the parasite-parasite interface in single hosts.
The health of western honey bees, Apis mellifera, has recently received considerable attention due to major losses of managed colonies at a global scale (Neumann & Carreck, 2010). Honey bee health is menaced by multiple stressors acting together or alone (Neumann & Carreck, 2010;Potts et al. 2010;Williams et al., 2010), with interactions among parasites likely to play a key role .
Whereas the two trypanosomatids, Crithidia mellificae and L. passim, can both infect honey bee colonies, L. passim is currently the predominant trypanosomatid in A. mellifera host populations globally . Based on accessioned sequences, all previous field data from honey bees in China, Italy, Japan, Spain, Switzerland, Turkey, and the USA were identified to be L. passim, and not C. mellificae as earlier suspected . Despite its global distribution, L. passim is poorly understood. However, it is known from laboratory experiments that mixed-species infections with C. mellificae and N. ceranae can significantly affect local and systemic immune gene transcription within honeybees . Such altered immune responses may also occur during mixed-species infections with L. passim and N. ceranae, but may differentially impact parasite populations.
As laboratory tests may not necessarily reflect field conditions (Retschnig et al., 2015), we took advantage of natural occurring infections of honey bee hosts with N. ceranae and L. passim in the field to test the following hypotheses: If the two parasites significantly interact with each other, we expect less or more individual hosts infected with both parasites compared to a random distribution. Likewise, infection loads of bees with one parasite alone should differ from those bees infected with both parasites. Lastly, positive or negative correlations between infection loads with both parasites are expected in doubleinfected hosts given that they interact with each other.

| Experimental setup
In April 2011, a total of 408 adult honey bee workers were sampled from the outer frames of four queenright A. mellifera colonies at each of three apiaries in the Swiss cantons of St. Gallen, Fribourg and Solothurn (N = 12 colonies total, Figure 2). All samples were transported on ice to the laboratory (Dainat, Evans, Chen, & Neumann, 2011;Human et al., 2013) and stored at −20°C until further analyses.

| DNA extraction
All sampled workers were crushed individually in 2-ml microcentrifuge tubes containing 5 mm metal beads and 200 μl TN buffer (10 mmol/L Tris, 10 mmol/L NaCl; pH 7.6). The samples were homogenized with a TissueLyser for 30 s at 25 1/s frequency using a Qiagen Retsch ® MM 300 mixer mill . Then, the homogenates were centrifuged at 2500 rpm and 50 μl of the supernatant was used for F I G U R E 1 Scanning electron microscope images of the two honey bee parasites: (a) Lotmaria passim; lanceolate promastigote cell with anterior flagellum and caudate posterior extension visible (×10,000), (b) Nosema ceranae: ovocylindrical, straight to slightly curved spores (×10,000). Bars = 2 μm

| Classical and quantitative PCR
PCR assays were performed to detect the occurrence of microsporidians (N. apis and N. ceranae) and trypanosomes (C. mellificae and L. passim). The PCR analyses were performed by using MyTaq™ kit (Bioline) with 2 μl tenfold-diluted DNA. We followed the manufacturer's recommendations by adding 5× reaction buffer, forward and reverse primers (final concentration of 0.4 μmol/L each) and 0.125 μl (0.63 Units) of Taq polymerase in a total of 25 μl final reaction volume. Three sets of species-specific primers available for Nosema species (multiplex PCR primer combination Mnapis-F, Mnceranae-F, and Mnuniv-R) and trypanosomes C. mellificae and L. passim (GAPDH-F/R and Cr ITS1-IR1/5.8R) were used (see Table 1). An independent PCR assay was performed for each set of primers. The qPCR cycling protocol was the same for each set of primers, which was as follows: cycling profile for all runs consisted of a 2-min initial incubation at

| Sequencing
Use of primers mentioned above requires subsequent sequencing of PCR products if the trypanosomatid species should be identified.
Therefore, PCR products from selected samples were sequenced using Cr-ITS1-IR1/Cr-ITS1-5.8R primers (Table 1). Additionally, selected PCR products from the N. ceranae assays were also sequenced using Mnceranae-F/Mnuniv-R primers (Table 1). N. ceranae and L. passim were confirmed as the parasites present in our samples using reference sequences deposited in GenBank.

| Data analyses
Lotmaria passim and N. ceranae parasites/bee were calculated fol- between L. passim and N. ceranae cell or spore equivalents was performed for those bees, which were infected by both parasites. We used a Chi-square test to assess for significant differences between the observed and expected frequencies. Significant differences in the proportions of infected individuals among apiaries were analyzed using simple logistic regression. All statistical analyses were performed using the program NCSS (NCSS 9 Statistical Analysis and Graphics).
The number of infected bees is significantly different from a random distribution, with N. ceranae being significantly more often found (Chisquare test, χ 2 = 4.324, df = 1, p < .05) than any other infection scenario. Furthermore, our data show no significant differences between observed and expected frequencies (i.e., bees infected with both parasites are not less common than expected from a random distribution) (  Table 4).
Overall, there were highly significant differences in infection levels (spore or cell equivalents/bee) between N. ceranae and L. passim

| DISCUSSION
The results of this study suggest little to no interaction between the two honey bee parasites L. passim and N. ceranae. This is possibly due to spatial separation in the naturally infected hosts (Higes, García-Palencia, Martín-Hernández, & Meana, 2007;Schwarz et al., 2015).
The observed positive correlation between L. passim and N. ceranae infection loads in double-infected workers most likely reflects differences among individual hosts, for example due to genetics, food and/ or age.
Neither C. mellificae nor N. apis were found in any of the analyzed workers. This confirms earlier studies that N. apis is currently rare in Switzerland (Dainat, Evans, Chen, Gauthier, & Neumann, 2012;Retschnig, Williams, & Neumann, in revision), and also supports that L. passim is currently the predominant trypanosomatid in A. mellifera F I G U R E 3 Infection levels of individual honeybee workers, Apis mellifera, with Nosema ceranae (N = 88), Lotmaria passim (N = 53) or both parasites (N = 8). Data are shown as spore or cell equivalents/ bee at a log-scale and display significant differences between N. ceranae and L. passim. No significant differences were found between bees infected with both parasites N. ceranae (& L. passim) and L. passim (& N. ceranae) and to single infected, respectively (means ± standard errors). All boxplots show the interquartile range (box), the median (black line within box), data range (horizontal black lines from box), and outliers (gray dots). Significant differences (Kruskal-Wallis ANOVA, Dunn's test, p < .0001) are indicated by letters (a,b) host populations . Conversely, infections with both L. passim and N. ceranae were equally common in all three apiaries, thereby confirming the widespread occurrence of these two honey bee parasites (Fries, 2010;Ravoet et al., 2013Ravoet et al., , 2015Schwarz et al., 2015). A higher number of workers infected with N. ceranae were found in Solothurn and St. Gallen compared to those from Fribourg.
Similarly, infections with L. passim were more often found in Fribourg compared to the other two locations. However, no significant differences were seen among the apiaries when individual bees were infected with both parasites. The underlying reasons for these observed differences in infections remain unclear.
As both L. passim and N. ceranae are common parasites of the honeybee , the absence of both parasites in the majority of samples seems interesting. However, one has to take into account that we did analyze individual workers in this study. Indeed, recent surveys using pooled worker bee samples showed that 46.7% of Swiss honey bee colonies are infected with Nosema spp. (Retschnig et al., in revision) and 82.5% with L. passim (Schneeberger, Yañez, Retschnig, Williams, & Neumann, 2017), respectively. Therefore, the overall prevalence of the two parasites in Switzerland is well in line with earlier studies for other regions (e.g., Schwarz et al., 2015;Stevanovic et al., 2011Stevanovic et al., , 2016. The data show no significant differences between the expected and observed frequencies of single-and double-infected bees. We therefore have no evidence that these two parasites interfere with each other's chances of infecting a host, pointing into the direction of no or weak parasite-parasite interactions. Similarly, a fairly high rate of honey bee colonies (60.5%) was coinfected with L. passim and N. ceranae over a 9 years survey, but no detectable correlation was found between the rates at which the two parasite-infected colonies (Stevanovic et al., 2016).
The observed infection loads of individual bees were significantly higher for L. passim compared to N. ceranae, and are well within the limits of previous reports for foraging bees (Bourgeois et al., 2010;Ravoet et al., 2013Ravoet et al., , 2015Retschnig, Neumann, & Williams, 2014;Retschnig et al., 2015). Despite the observed high infection loads, there were no significant differences between the infection loads of both parasites in single-or double-infected hosts. Again, this provides no support for either competition or cooperation between the two parasites, even though single-and mixed-species C. mellificae trypanosome and N. ceranae microsporidia infections elicit distinct immune responses .
So, why are there no significant interactions between the two intestine parasites N. ceranae and L. passim despite high infection loads? While N. ceranae has been detected using PCR in other tissues than the gut (i.e., hypopharyngeal glands, salivary glands, Malpighian tubules, and fat body, Chen et al., 2009), this microsporidian only develops spores in the midgut (Higes et al., 2007). However, L. passim prefers to colonize the rectum tissue of honey bees . This spatial separation in infected hosts may explain why the interactions between the two parasites are not significant, because they actually might not compete for the same niche. However, on the other hand, the significant positive correlation between L. passim and N. ceranae infection loads in concurrently infected hosts may indicate some cooperation between the parasites (Alizon & Lion, 2011).
Genetically based differences in honey bee susceptibility to pathogens are long known (Rothenbuhler, 1964) and have also been found for N. ceranae (Huang et al., 2014). Moreover, honey bee colonies consist of many different subfamilies (patrilines) due to high degrees of polyandry of the queens (Neumann, Moritz, & van Praagh, 1999). It is therefore extremely likely that honey bee workers even from the very same colony can differ in their susceptibility to parasites due to genetics alone, thereby offering an alternative explanation for the observed positive correlation between infection loads given that some hosts are more vulnerable to both parasites compared to others (i.e., higher or lower infection loads for both parasites, respectively). Alternatively, environmentally imposed differences among individual hosts can also explain this trend.
For example, protein fed to honey bee workers can increase N. ceranae spore numbers (Jack et al., 2016), and some workers may have consumed more pollen than others. Finally, it is known that age can play a key role for N. ceranae spore loads of honey bees (Smart & Sheppard, 2012). Therefore, the observed significant correlation between infections loads of the two parasites L. passim and N. ceranae can be parsimoniously explained without assuming any interactions whatsoever between the two parasites. In light of our other results and our random sampling of bees, we consider it more likely that differences among hosts are underlying the observed positive correlation between L. passim and N. ceranae infection loads. In any case, correlation is obviously not causation, and further experiments with bees of both known age and pollen consumption under single and double infection scenarios are required to understand the underlying mechanisms for the observed correlation.
F I G U R E 4 Infection levels of individual honeybee workers, Apis mellifera, with both Nosema ceranae and Lotmaria passim. A significant positive correlation was found (n = 8, Pearson |r| = .81, p = .015) In conclusion, our data do not suggest any strong interactions between the two honey bee parasites, which may be explained by spatial separation in the host.