Many pollinator species are declining due to a variety of interacting stressors including pathogens, sparking interest in understanding factors that could mitigate these outcomes. Diet can affect host-pathogen interactions by changing nutritional reserves or providing bioactive secondary chemicals. Recent work found that sunflower pollen (Helianthus annuus) dramatically reduced cell counts of the gut pathogen Crithidia bombi in bumble bee workers (Bombus impatiens), but the mechanism underlying this effect is unknown. Here we analyzed methanolic extracts of sunflower pollen by LC-MS and identified triscoumaroyl spermidines as the major secondary metabolite components, along with a flavonoid quercetin-3-O-hexoside and a quercetin-3-O-(6-O-malonyl)-hexoside. We then tested the effect of triscoumaroyl spermidine and rutin (as a proxy for quercetin glycosides) on Crithidia infection in B. impatiens, compared to buckwheat pollen (Fagopyrum esculentum) as a negative control and sunflower pollen as a positive control. In addition, we tested the effect of nine fatty acids from sunflower pollen individually and in combination using similar methods. Although sunflower pollen consistently reduced Crithidia relative to control pollen, none of the compounds we tested had significant effects. In addition, diet treatments did not affect mortality, or sucrose or pollen consumption. Thus, the mechanisms underlying the medicinal effect of sunflower are still unknown; future work could use bioactivity-guided fractionation to more efficiently target compounds of interest, and explore non-chemical mechanisms. Ultimately, identifying the mechanism underlying the effect of sunflower pollen on pathogens will open up new avenues for managing bee health.

The general approach was similar for all assays and followed previously published protocols (Giacomini et al. 2018; LoCascio et al. 2019a; Richardson et al. 2015). Commercial experimental colonies (BioBest, Leamington, Canada) were confirmed to be Crithidia-free by biweekly subsampling, and the Crithidia strain used in these assays was originally collected in 2015 from infected wild B. impatiens at a farm in Hadley, Massachusetts USA (42°21'51.93"N, 72°33'55.88"W) and maintained in commercial colonies thereafter. All colonies were fed mixed wildflower pollen (Koppert Biological Systems, Howell, Michigan, USA or CC Pollen Inc., Phoenix, Arizona, USA). Worker bees were inoculated after a 1-2 hr starving period with either 10 µl (secondary compounds) or 15 µl (fatty acids) of a 25% sucrose solution containing 600 cells/µl, well within the range of natural fecal Crithidia concentrations (Otterstatter and Thomson 2006). Upon inoculation, bees were placed in individual 18.5 ml vials (secondary compounds) or Placon 473 ml cups (fatty acids) and assigned to diet treatments that they received for 7 days along with 30% sucrose solution that was replaced daily (secondary compounds) or every other day (fatty acids). Bees were maintained in darkness at 27^oC in an incubator during assays. After 7 days, bees were dissected to remove the gut, which was ground in 300 µl of ¼ strength Ringer’s solution (Sigma-Aldrich – Fluka 96724). After 4 hr, a 10 µl subsample was removed and moving Crithidia cells were counted on a 0.02 µl field of vision on the grid of a hemacytometer.     

For the secondary compound assays we used callows, which are bees that had emerged from pupae within the last 24 hr. We isolated pupal clumps from colonies, collected callows as they emerged and fed them wildflower pollen for two days before inoculating them to enter the experiment. Because callows emerged without nestmates, this was an extra precaution to ensure they were free of Crithidia (which infects adults), but these bees also did not receive gut microbiota from their nestmates that could influence interactions with Crithidia (Koch and Schmid-Hempel 2011). For the fatty acid assays, we pulled adult workers directly from colonies and inoculated them to enter the experiment. This method does not control for worker age, but allows bees to acquire their colony’s gut microbiome. Within each assay, bees with pollen treatments were compared to controls under the same conditions, so while our methods varied, we can still assess diet effects within each experiment. We also included a measure of bee size for all assays since smaller bees often have higher infections (e. g., Richardson et al. 2015). For the secondary compound assays we used callow mass, and for fatty acids we used radial cell length, which is highly correlated with other measures of body size (Nooten and Rehan 2020). In the triscoumaryl spermidine experiment, we measured both callow mass and radial cell length and they were highly correlated (Pearson’s r = 0.85, n = 259). 

For all assays we recorded daily mortality, and measured daily pollen and sucrose consumption once per bee, except for omitting sucrose consumption in the triscoumaroyl spermidine assay. Pollen and sucrose rations were typically weighed upon providing them to bees and then approximately 24 hr later (48 hr in the first fatty acid assay).

Secondary Chemical Bioassays. We tested the effects of two major secondary compounds found in sunflower pollen, a triscoumaroyl spermidine and rutin, at four concentrations each. We used rutin as a proxy for the quercetin glycosides identified in sunflower pollen because rutin is also a quercetin glycoside and available commercially. Quercetin glycosides are reported to show equivalent activity across a range of analogues against microorganisms including trypanosomes (da Silva et al. 2019; Marin et al. 2017), suggesting that rutin provides a suitable proxy for other quercetin glycosides.

Tri-p-coumaroyl spermidine (‘triscoumaryl spermidine’ hereafter) was synthesized in the laboratory as described in Online Resource 1, while rutin was purchased commercially (Sigma-Aldrich, R5143-50G). Both were added to buckwheat pollen at 1% (very high), 0.1% (high), 0.01% (medium), or 0.001% (low) concentrations wt/wt. For context, we found 12 mg/g of triscoumaroyl spermidine in pollen (1.2%; comparable with the ‘very high’ treatment) and 0.2% quercetin glycosides in pollen, similar to the ‘high’ level. Thus, our concentrations are within natural ranges. Compounds were mixed with buckwheat pollen and compared to pure buckwheat pollen as a negative control and sunflower pollen as a positive control (sunflower and buckwheat pollen came from Changge Huading Wax Industry Co., Ltd., Henan, China; except that buckwheat came from Fuyang Import and Export Ltd for the final fatty acid assay). Compounds were added to dry pollen, which was then mixed with distilled water for all treatments to create a paste that was frozen at -20^oC until use. Details of mixing secondary compounds with pollen and preparing pollen balls are in Online Resource 2.

The triscoumaryl spermidine assay was conducted from July 17 through August 19, 2015, using 108 bees from three experimental colonies, ultimately including 15-20 bees/treatment. The rutin experiment was conducted from November 3, 2015 through April 7, 2016 and used 292 bees from 12 experimental colonies, ultimately including 45-54 bees/treatment.

Fatty Acid Bioassays. We tested the effects of several fatty acids singly and in combination. We chose the most common ones detected in sunflower pollen that were commercially available, and tested each at typical concentrations based on published literature (Nicolson and Human 2013; Yang et al. 2013). 

We conducted three assays that each tested individual or combinations of fatty acids added directly to control pollen as powders or liquids. Compounds tested in the first assay were linoleic (1.67 mg/g; CAS 60-33-3), lauric (12.39 mg/g; CAS 143-07-7), caprylic (1.46 mg/g; CAS 124-07-2), palmitic (8.46 mg/g; CAS 57-10-3), and decanoic (0.14 mg/g; CA: 334-48-5) acids. The second assay included linolenic (7.61 mg/g; CAS 463-40-1), myristic (1.62 mg/g; CAS 544-63-8), stearic (0.65 mg/g; CAS 57-11-4) and oleic (5.39 mg/g; CAS 112-80-1) acids. The third assay tested all nine fatty acids combined in a single treatment (using the same concentrations as in individual assays) and sunflower oil, which contains a mix of fatty acids (37.07 mg/g; Organic 365 Everyday Value Products, Whole Foods Market, Austin, Texas, USA), to assess whether fatty acids had interactive effects that would not be detected in assays of individual compounds. Details of suppliers and concentration calculations are provided in Online Resource 2. Fatty acids were added to dry pollen with a coffee grinder, and then distilled water was added in a 6:1 or 7:1 (third assay only) ratio to make a paste. As with the secondary chemical assays, compounds were added to buckwheat pollen and compared to buckwheat and sunflower pollen as negative and positive controls, respectively. However, in the second assay we did not have sufficient buckwheat pollen and instead added fatty acids to mixed wildflower pollen (CC Pollen Inc. Phoenix, Arizona, USA) and used wildflower pollen as the negative control.

            We conducted the first assay from September 28, 2018 to February 6, 2019 using 246 bees from 14 experimental colonies, including 32-37 bees per fatty acid treatments and 47 in the buckwheat control. The second was conducted from February 20 to April 12, 2019, using 167 bees from 5 experimental colonies, including 25-31 bees/treatment. The third assay was conducted from July 18 to 26, 2019 using 106 bees from 4 experimental colonies, including 25-27 bees/treatment.

All variables for all CSV files are described in the 'ReadMe' text file.

A single R script was used to analyze all data and is also uploaded. 

There are missing values in the Crithidia count datasets when bees died before the 7-day span of the bioassays.