Protein nutrition governs within-host race of honey bee pathogens

Multiple infections are common in honey bees, Apis mellifera, but the possible role of nutrition in this regard is poorly understood. Microsporidian infections, which are promoted by protein-fed, can negatively correlate with virus infections, but the role of protein nutrition for the microsporidian-virus interface is unknown. Here, we challenged naturally deformed wing virus - B (DWV-B) infected adult honey bee workers fed with or without pollen ( = protein) in hoarding cages, with the microsporidian Nosema ceranae. Bee mortality was recorded for 14 days and N. ceranae spore loads and DWV-B titers were quantified. Amongst the groups inoculated with N. ceranae, more spores were counted in protein-fed bees. However, N. ceranae infected bees without protein-diet had reduced longevity compared to all other groups. N. ceranae infection had no effect on protein-fed bee’s longevity, whereas bees supplied only with sugar-water showed reduced survival. Our data also support that protein-feeding can have a significant negative impact on virus infections in insects. The negative correlation between N. ceranae spore loads and DWV-B titers was stronger expressed in protein-fed hosts. Proteins not only enhance survival of infected hosts, but also significantly shape the microsporidian-virus interface, probably due to increased spore production and enhanced host immunity.

and Nosema apis 28 , N. ceranae and DWV 29,30 . Synergistic effects have been reported between N. apis and several viruses, e.g. filamentous virus, bee virus Y and black queen cell virus (BQCV) 31,32 . In contrast, no association was found between Israel acute paralysis virus (IAPV) and Nosema ceranae 33 . This large range of possible interactions between pathogens in one host calls for investigation of possible mechanisms driving this interface.
Since it is known that proteins can impact both virus and microsporidian infections in insects, we regard it as likely that this will affect the outcome of virus-microsporidian interactions in multiple infected hosts. Pollen is the main natural source of protein for honey bees, especially for young bees. It is an essential protein source and may interfere with pathogen-pathogen interactions, infection intensity and longevity of the honey bee host 34,35 .
Here, we investigated the possible role of protein feed via pollen on the interface between DWV-B (formerly Varroa destructor virus-1) and Nosema ceranae in individual honey bee workers. We hypothesize that protein fed in the form of pollen will not only have significant beneficial effects for the hosts, but will also amplify the virus-microsporidian interface.

Results
Since the experimental pollen was not irradiated, some bees (N = 19) of the Pollen-only treatment were naturally infected with N. ceranae. These contaminated bees were excluded from further data analyses.
The virus strain-specific PCR 36 showed that only DWV-B was infesting the experimental bees. Neither BQCV, DWV, nor acute bee paralysis virus (ABPV) were found in any of the analyzed individual honey bee workers (N = 120).
All data (N. ceranae spore loads, DWV-B infection levels, sugar and pollen consumption) were not normally distributed (Shapiro-Wilk's test for normality, P < 0.05 in all cases). Therefore, the non-parametric Kruskal-Wallis multiple comparisons One Way ANOVA, Dunn's Tests were performed.
The survival in the different groups is shown in (Fig. 1). While a significantly reduced longevity was observed for Nosema-only infected bees compared to all other groups (Kaplan-Meier, Log-Rank test, P < 0.05), N. ceranae infection had no significant effect on the longevity of pollen-fed bees (Kaplan-Meier, Log-Rank test, P > 0.05). Bees that were supplied only with sugar water showed a reduced survival compared to the bees which received both sugar and pollen (Kaplan-Meier, Log-Rank test, P < 0.05, Fig. 1). Hence, workers exposed to both N. ceranae and pollen showed a non-additive effect when compared to both treatments individually (χ 2 = 22.73 (equation-3), theoretical χ 2 = 7.879, df = 1, P = 0.005). Due to the calculated negative value −33.87 (equation-4), the observed effect on mortality can be considered antagonistic.
The correlation between N. ceranae spore loads and DWV-B infection levels was not significant in the Nosema-only treatment (Pearson Correlation: Pearson |r| = −0.22, P = 0.12; Fig. 4). However, a significant  ). Medians, ranges, confidence intervals and outliers ( = dots) are shown at a log scale. Significant differences were found between the groups Nosema-pollen and Nosema-only, as well as between the two Nosema groups and the Controls. Please note that N = 19 bees in the Pollen-only treatment were naturally contaminated with N. ceranae. When excluding these contaminated bees from the Pollen-only group, significant differences were still found between the groups Nosema-pollen and Nosema-only (*P = 0.025, **P < 0.0001). The two pollen-fed treatments (Pollen-only, Nosema-pollen) showed no significant difference in pollen consumption over the 14 days (Kruskal-Wallis multiple comparisons One Way ANOVA, Dunn's Test z > 1.96, P > 0.05, Fig. 6).

Discussion
Our data show for the first time that protein-feeding can have a significant impact on the microsporidian-virus interface in double-infected insect hosts. Taken together with the here confirmed impact on microsporidian 8 and virus 5,37 infections and on host survival, our results also provide strong support that protein nutrition can functionally alter not only general resistance in insects, but also alter the pattern of host-parasite interactions 9 . The experimental pollen was not sterilized similar to other studies 8 . Therefore, some of the Pollen-only treated workers in our experiment became naturally infected with N. ceranae. These bees were excluded from any further data analyses. Due to possibly associated probiotic microorganisms 38 , such non-irradiated pollen is probably more beneficial than irradiated one and thus more likely to reveal the full potential of adequate pollen-fed  for honey bee resilience. It also appears likely that bees from the Nosema-pollen treatment have obtained additional spores. However, this seems not relevant due to the very high number of N. ceranae spores used for the treatments.
Our data confirm that Nosema spp. infected honey bee workers display higher spore loads when they are pollen-fed (N. apis: 6 ; N. ceranae: 7,39,40 ). This is most likely because Nosema spp. are highly dependent on host nutritional status for their own development, e.g. host amino acids 41,42 , Adenosine triphosphate (ATP) 43,44 or other core nutrients 45 . Therefore, it appears obvious that any individual host with a pollen-rich diet becomes ideal for Nosema spp. reproduction simply by providing a supreme nutritional environment. However, since protein fed is likely to enhance the immune system 5,9 , the question emerges, why the bees were not able to reduce the N. ceranae spore load. This might be explained by the lack of host-parasite co-evolution between A. mellifera and the fairly recent invasive species N. ceranae. This scenario seems likely because beekeepers limit natural selection, thereby preventing adaptation of honey bees to this and other novel parasites (i.e. Varroa destructor) 46 . Alternatively, but not mutually exclusive, N. ceranae might have interfered with the host immune response 47 . Nevertheless, despite higher spore loads these N. ceranae infected individuals showed an improved survival confirming earlier findings 8 . Adequate pollen availability might compensate for the energy and nutrients lost in honey bees with high N. ceranae infection intensity, thereby enabling improved survival 8 . These results are also in line with another study 48 that nutrition influences survival in colony level N. ceranae infections. In a simultaneous choice test between sunflower honeys and honeydew N. ceranae infected bees significantly preferred sunflower honey over honey dew 49 Such bees consuming sunflower honey showed significantly lower N. ceranae spore loads compared to the honeydew group, probably because of the higher antibiotic activity of sun flower honey 49 . These findings are well in line with therapeutic self-medication reported from primates 50 and butterflies 51 .
Our data also provide support to the key role of protein nutrition for the outcome of virus infections in insect hosts 5,37 . Indeed, bees only receiving sugar water showed a significantly higher mortality and higher DWV-B infection levels compared to pollen-fed ones. This confirms that DWV infected bees fed with pollen show lower viral loads than bees only fed with sugar water 52 . Since just a subsample of 30 bees per treatment was tested for other viruses (BQCV, DWV, ABPV), it can obviously not be excluded that those viruses might also have contributed to a higher bee mortality. However, none of the other viruses were found in any of the analyzed honey bees suggesting a high probability for their absence.
Most interestingly, the results show for the first time that protein-feeding can significantly impact the microsporidian-virus interface in double-infected insect hosts. What are possible reasons for this shifted pathogen interface? In line with other studies 30 , pre-infection of DWV did not interfere with N. ceranae replication, but N. ceranae did interfere with DWV replication. Since N. ceranae replication in midgut cells disrupts protein metabolism and causes energetic stress 8,53,54 , this microsporidian is likely to compete with orally acquired viruses for cell resources or for DWV-B's accessibility to midgut cells 30 . This could explain why an increase in cell resources due to better protein fed led to higher N. ceranae spore counts, but also resulted in lower virus loads. Moreover, N. ceranae was shown to induce a significant increase in phenol oxidase in bees fed on sugar and workers' longevity in cages was positively linked to phenol oxidase activity 55 . This suggests a second possible explanation for the stronger negative correlation between N. ceranae and DWV-B in protein-fed hosts because both phenol oxidase and upregulation of antimicrobial peptides are linked with effective antiviral responses in A. mellifera 56 . Since pollen promotes development of the main honey bee immune organ, the fat body 57 , its maturation may enable a better performance of the immune system including activation of phenol oxidase. Indeed, pollen diet promotes fat body development and enhances survival of N. ceranae parasitized workers that have expressed higher levels of vitellogenin and immunoprotein 55,58-61 . Moreover, it is known that pollen supplies are in general immensely important for overwintering honey bee colonies to effectively oppose pathogen stress 40 . Interestingly, Spaetzle, an activator of the Toll pathway was upregulated by pollen feeding in healthy bees as well as a gene coding for the antimicrobial peptide Defensin and the Peptidoglycan recognition protein PGRP-LC 62 . In conclusion, the conjunction of increased N. ceranae spore production and enhanced host immunity acquired by proficient protein nutrition most likely explains the observed significant increase in the negative correlation between N. ceranae and DWV-B loads. Alternatively, but not mutually exclusive, the negative correlations between N. ceranae and DWV infections as found in this study and in others 29,30,63 can also be explained by individual workers being more susceptible to microsporidians, but less susceptible to virus infections. For example, some bees may be more susceptible to Nosema spp. infections, but could be better able to resist viruses resulting in an overall significant correlation. Indeed, a genetic basis for disease susceptibility is long known in honey bees 64 , incl. Nosema spp. infections 65 , and colonies consist of a large number of subfamilies, so-called patrilines, due to the high degree of polyandry by queens 66 . This provides an alternative explanation for the observed negative correlation between the two pathogens in individual hosts without assuming any antagonistic interactions between them.
No differences in the amount of consumed sugar water were found between control and the Pollen-only group compared to the two N. ceranae -treated groups similar to an earlier study 8 . The higher amounts of sugar-water consumption by N. ceranae-infected bees observed in other studies 67-69 may be linked to the lack of protein fed in the respective experimental designs. Our study also revealed a significantly higher sugar water consumption in the Nosema-only group compared to Pollen-only workers, suggesting that lack of protein combined with microsporidian infections may result in higher hunger levels 67 . In line with an earlier study 8 , we found no significant differences in the amount of pollen consumed between the Pollen-only and the Nosema-pollen groups. Even though N. ceranae infections certainly induces costs for the hosts, those do apparently not result in significant differences in protein consumption between infected and non-infected bees at least over our experimental period. In conclusion, irrespective of the actual mechanisms underlying the observed stronger negative correlation between microsporidian and virus infection levels in protein-fed hosts, our data strongly suggest that proteins can govern the pathogen-pathogen interface in double-infected insect hosts. Our results further provide support that protein nutrition is an overall key factor for the outcome of infections in insects [5][6][7][8][9] .

Material and Methods
Study design. The experiment was conducted in September 2014 at the Institute of Zoology, Karl-Franzens University, Graz, Austria using honey bee workers from four randomly chosen queenright local colonies (predominantly A. m. carnica). All colonies were routinely treated against V. destructor in late summer using formic acid and oxalic acid in the previous winter [70][71][72] .
To test whether pollen nutrition has an effect on N. ceranae interactions with naturally occurring virus infections, a fully-crossed hoarding cage experiment was performed with four replicates each: 1. Workers fed with sugar, but not with pollen (=Controls); 2. Workers fed with both sugar and pollen (=Pollen-only); 3. Workers fed with N. ceranae spores and sugar, but without pollen (=Nosema-only); 4. Workers fed with N. ceranae spores, sugar and pollen (=Nosema-pollen). Spore solutions. The N. ceranae spore solutions were prepared following routine protocols 73 . In brief, 12 foragers were collected from the hive entrances of four local infested colonies and dissected. Then, three midguts of N. ceranae infested workers were pooled together in a vial with 0.5 ml water. After homogenization, each vial solution was checked under the light microscope (x400 objective) for the presence of N. ceranae spores 74 . After all four vials were checked for positive spore loads; the spore solutions were mixed and centrifuged at 5 000 rpm for 5 min. The supernatant containing tissue debris was discarded and the spore pellet was re-suspended in 0.5 ml water by vortexing for 5 sec until spores were uniformly distributed in the solution. This washing step was repeated twice until the N. ceranae spore solution had a concentration of at least 85% purity 73 . The re-suspended solution was 500 µl water in a 1.5 ml Eppendorf ® tube prior to spore load quantification using a haemocytometer and light microscopy (Thermo Fisher Scientific, Waltham, Massachusetts, USA) 75,76 focusing on five large squares (each containing 16 small squares) in which the N. ceranae spores were counted. The final concentrations of the spore solutions were quantified using the following calculation 77 : where S N is the number of spores per honey bee in 500 µl and S H is the number of spores in 5 large haemocytometer squares (80 small squares). The taxonomic status of the spores was confirmed using N. ceranae species-specific PCR 73 for 30 individual honey bees of each treatment.
Experimental set up. Four frames with sealed worker brood were taken from each of the four experimental colonies and placed in an incubator at (34.5 °C) until adult emergence. To ensure that the bees were not older than 24 h, all bees on the brood frames were removed the evening before the experiment started. Each treatment group consisted of six standard hoarding cages with 50 workers randomly assigned to each cage 78 . Bees were fed with 50% sugar water (w/v) ad libitum until the 3 rd day 28 , at which pollen feeding and N. ceranae infection started.
Prior to the treatment, all bees were starved for 2 h 29,53,75 before N. ceranae infection was done by bulk feeding 73 over 24 h. Workers of the two N. ceranae treatments were challenged with ∼100'000 spores per bee 75,78-80 . All workers were fed until the end of the experiment (14 days) ad libitum with 50% sugar water (w/v). In addition, pollen-treated bees were provided with pollen dough containing corbicula pollen and sucrose candy 78 . Prior to the experiments, this pollen was not gamma ray irradiated. The experiment lasted 14 d, in which the honey bees were kept in an incubator at brood nest temperature (34.5 °C) 81 with 75% RH 82 , for the first 6 days, before the temperature was decreased to 30 °C 78 for the remaining experimental period of 8 days.
To test for potential differences in nutritional demand, pollen dough and sugar water consumption was measured in all cages on a daily basis 67 . The syringes prepared as feeders were refilled every other day in order to avoid the formation of mold 78 . Dead bees were removed daily 76,83 . Possible cage effects were expected as random effects, whereas all replicates had the same conditions in the incubator, temperature and random mixture of bees.

Virus infection levels.
All 401 bees were individually analyzed for DWV-B. Prior to RNA extraction, individual bees were crushed in 2 ml microcentrifuge tubes containing 5 mm metal beads and 200 µl TN-buffer (10 mM Tris, 10 mM NaCl; pH 7.6). The samples were homogenized with a tissue-lyser for 30 sec at 25 1/s frequency using a Qiagen Retsch ® MM 300 mixer mill 84 . Then, the homogenates were centrifuged at 2500 rpm and 50 µl of the supernatant was destined for total RNA extraction using NucleoSpin RNA extraction kit (Macherey-Nagel) following the manufactures guidelines. Reverse transcription was performed using 2 µg of extracted RNA incubated with random hexamer primers for 5 min at 70 °C. Then mixed with 5 µl of 5x buffer, 1.25 µl dNTP (10 mM) and 1 µl M-MLV before incubating the 25 µl volume reaction for 60 min at 37 °C. For virus quantification, 10-fold diluted cDNA was mixed with Kapa SYBR ® FAST qPCR Master Mix kit. Briefly, 6 µl 2x reaction buffer, 0.24 µl forward and reverse primers for DWV-B and β-actin (Table 1) 85 merged with 2.52 µl water and 3 µl template in a total of 12 µl final reaction volume 86 . The real-time qPCR cycling profile consisted of 3 min incubation at 95 °C and 40 cycles of 3 sec at 95 °C for denaturation, 30 sec at 57 °C for annealing and data collection. The melting-curve analysis was performed with the following conditions: 15 s at 95 °C, 55 °C and 95 °C, respectively. Purified DWV-B PCR products of know concentration (10 −3 -10 −6 ng) were used as standard curves on each individual plate, along with non-template controls (R 2 : 0.992; Slope: −3.198; Intercept: 30.240; PCR efficiency: 2.054). Quantification of the β-actin gene was performed in parallel for each sample as reference gene for DWV-B normalization 85 . A Cq cut-off value (according to the value of the negative control) was used to define the disease status (positive or negative). The ECO Software real-time PCR system (Illumina, San Diego, CA, USA) was used to evaluate the performance of the qPCR reactions and to analyze the qPCR quantification. These Cq-values were used to calculate the virus infection levels in Log [copies/bees] which were then used for the statistical analyses.
A further subsample of 30 bees per treatment (N = 5 each cage) was screened to determine if other honey bee viruses were also present in the colonies during the experiment: BQCV, which is known to be associated with N apis infection 31,32 and occurs in 30% of Austrian honey bees 87 . Three more viruses associated with honey bees in Austria were screened: DWV which is present in 91% of Austrian honey bees including its variant DWV-B and ABPV reported to be present in 68% of Austrian bees 87 .
Longevity analyses for the four individual treatment groups were conducted by using Kaplan-Meier Survival Curves and a Log-Rank assessment.
Interactions between treatments on worker mortality were determined by using χ 2 tests 88,89 . The expected interaction mortality value, M E for combined treatment was calculated using the following formula: where M O is the observed mortality for the combined N. ceranae with pollen treatment. A non-additive effect between the two agents was expected when the χ 2 value exceeded the given table value. If, however the difference between had a positive or a negative value, an interaction was then regarded as being synergistic or antagonistic, respectively 88 . Synergistic, additive, or antagonistic interactions between agents in the combination treatments for DWV-B loads and N. ceranae spores were determined using a χ 2 test 90-93 . Comparisons of N. ceranae spore loads, DWV-B infection levels and sugar/pollen consumption rates in the different treatment groups were performed using Kruskal-Wallis One Way ANOVAs (and multiple comparisons, Dunn's Test).
Pearson correlations between N. ceranae spore loads and DWV-B infections levels (Log [copies/bee]) were performed for both Nosema-only and Nosema-pollen groups.
All statistical analyses were performed using the program NCSS (NCSS 9 Statistical Analysis and Graphics). CTT CCT CAT TAA CTG AGT TGT TGT C Table 1. Primers used for the quantification of honey bee viruses by qPCR assays. The targets, primer names, sequences, the product size and references are shown.