Fatty acids and their derivatives from Chlorella vulgaris extracts exhibit in vitro antimicrobial activity against the honey bee pathogen Paenibacillus larvae

Abstract The green microalga Chlorella vulgaris Beijerinck (Chlorellaceae) is widely used as a food supplement for humans and animals. In beekeeping practice, Chlorella vulgaris has potential as a pollen supplement. We studied whether Chlorella extracts display antimicrobial properties against Paenibacillus larvae, the causative agent of the honey bee bacterial disease American foulbrood. We identified components responsible for antimicrobial activity and evaluated the added values of Chlorella as a food supplement for honey bees. Different extracts (water, acetone, methanol) were prepared from Chlorella biomass (phototrophically and heterotrophically cultivated) and screened for antimicrobial activity against ERIC I and ERIC II genotypes of P. larvae. Active acetone extracts of phototrophically cultivated Chlorella vulgaris biomass were fractioned via preparative reverse-phase chromatography. Antimicrobial activity was detected for 9 of the resulting 33 fractions. Further analysis revealed the chemical composition of the active fractions. C. vulgaris extracts showed a significant antimicrobial effect against vegetative cells and spores of P. larvae strains of ERIC I and ERIC II genotypes. The lowest MIC of the most active acetone extract was 6.3 µg/mL for both tested genotypes. In the majority of the active fractions, monolinolenin, fatty acid linoleic acid, and methyl esters of linoleic and/or palmitic acid were identified via high-performance liquid chromatography coupled with high-resolution mass spectrometry analysis. Based on our results, we concluded that algal C. vulgaris food supplements not only contain nutritional but also potential prophylactic properties for honey bee health.


Introduction
The spore-forming bacterium Paenibacillus larvae is one of the most serious honey bee (Apis mellifera) pathogens and the causative agent of a globally distributed and highly contagious disease named American foulbrood (AFB) (Genersch, 2010).Usually, colonies with clinical sights of AFB have to be destroyed to prevent the disease from spreading to the neighbouring colonies, which leads to a high economic impact on bee farms (Smith et al., 2013;Vanengelsdorp & Meixner, 2010).AFB affects the larvae of honey bees, whereas adult bees only transmit the infectious spores as asymptomatic carriers.P. larvae spores are the sole infectious form of this organism.The most susceptible to the P. larvae spores are the youngest bee larvae (12-36 h after hatching) (Genersch, 2010).Once the larvae consume food sources contaminated with P. larvae endospores, the spores germinate in the larval midgut.After several days, the development of the infection results in the death of the larvae, which decomposes into a brownish glue-like fluid containing millions of P. larvae spores (Genersch, 2010).
To control AFB, certain countries use antibiotics (oxytetracycline and tylosin), which suppress the clinical phase of the disease but do not affect P. larvae spores.Antibiotic treatment, however, is not allowed in European Union countries owing to residual contamination of beehive products, especially honey (Mutinelli, 2003).Therefore, the affected colonies are usually destroyed by burning the hives.Due to a lack of approaches to effectively control and prevent bee colonies from developing AFB, several studies have investigated natural substances showing antimicrobial activity against P. larvae (Alonso- Salces et al., 2017;Chaimanee et al., 2017;Flesar et al., 2010;€ Ozkirim et al., 2012).Such compounds may be administered as a prophylactic feed to honey bees during unfavourable periods to decrease the risk of disease outbreaks (Doull, 1975).
Microalgae represent a rich natural source of biologically active compounds.Catarina Guedes et al. (2011) described the antimicrobial activity of water and ethanol extracts derived from 11 species of microalgae against important food pathogens, e.g., Salmonella sp. or Escherichia coli.Salem (2014) discovered the antimicrobial and antifungal effects of acetone and methanol extracts of several different microalgae.Moreover, other previously identified compounds may show antiviral, antimicrobial, antiinflammatory, or antioxidant activity (de Morais et al., 2015).For instance, pheophorbide a, a degradation product of chlorophyll a, shows antiviral and antimicrobial activity (Ratnoglik et al., 2014;Villasenor & Carino, 2011).Due to their biomass composition, microalgae are used as a food supplement in the human diet and also as a feed supplement for animals (Bai et al., 2001;Lamminen et al., 2019).
C. vulgaris biomass has been previously shown to be well consumed by honey bees with a positive effect on physiological parameters and colony growth (Eremia et al., 2013;Jehl ık et al., 2019;Ricigliano & Simone-Finstrom, 2020).However, the antimicrobial effect of C. vulgaris biomass against P. larvae has not yet been studied.Therefore, in this study, we analysed the antimicrobial effects of C. vulgaris extracts on P. larvae vegetative cells and spores to determine whether Chlorella biomass or extract may be used as a feed supplement with a prophylactic function against bacterial diseases in bees.

Alga cultivation and degradation
The phototrophically grown biomass (CP) of the fresh water microalgae Chlorella vulgaris R-117 (CCALA 1107, Culture Collection of 81 Autotrophic Organisms, Institute of Botany, T rebo n, Czech Republic) was cultivated using outdoor cultivation units (Masojidek et al., 2011).Heterotrophically grown culture (CH) was cultivated according to Doucha and L ıvansk y (2011).The fresh harvested biomass was spray dried at 220 C. Degraded C. vulgaris biomass (CPD) was prepared from 10 g of phototrophically grown biomass which was spread on a plate and placed next to a window for two months until its colour turned grey.The biomass was protected from contamination by covering the plate with translucent foil.The powder was mixed up once a week.

Water and methanol extracts
A sample (10 g) of dried algal biomass was homogenized with the sea sand (Lach-Ner s.r.o., Czech Republic) using mortar and pestle in 100 mL of deionized water or 70% methanol (v/v) (Sigma Aldrich, USA) and after 1 h the suspension was centrifuged for 10 min at 1920 Â g (Eppendorf, Germany).The supernatant was freeze-dried (water extracts) or evaporated until dryness using a rotary evaporator (methanol extracts) (Eppendorf, Germany).

Acetone extracts
A sample (10 g) of dried algal biomass was homogenized with the sea sand (Lach-Ner s.r.o, Czech Republic) using a mortar and pestle in 10 mL of 100% acetone (Sigma Aldrich, USA).After centrifugation (Eppendorf, Germany, 10 min, 1920 Â g) the supernatant was collected, and the sediment was reextracted with 10 mL volumes of acetone until the colourless pellet was obtained.n-Hexane (Sigma Aldrich, USA) was added to the pooled supernatants (volume of n-hexane was 1 = 2 of the total volume of acetone used for extraction) and the mixture was shaken.The same volume of water identical to nhexane volume was added into the mixture and mixed carefully.The upper layer was removed and dried under a stream of nitrogen.

Fractionation of extract
Extract for fractionation was prepared from 10 g of algal biomass (CP) extracted with 200 mL of 60% acetone.After centrifugation, 42 mL of crude extract was mixed with 40 mL of acetone, 40 mL of n-hexane, and 15 mL of water.The upper layer was dried in rotary evaporator at 30 C, the residue was re-suspended in small amount of methanol to make saturated solution and subjected to a chromatographic separation using an HPLC Agilent 1100 system equipped with a DAD detector.The separation was performed using a semipreparative column Agilent Zorbax Eclipse XDB-C18 (9.4 Â 250 mm, 5 mm).Mobile phases were methanol (A) and water (B), both containing 0.1% formic acid (Sigma Aldrich, USA).Gradient for the separation was as follows: 0 min 50% A, 5 min 90% A, from 15 to 40 min 100% A, 42 min 50% A and 47 min 50% A. Fractions were collected from 5 to 15 min every two minutes, and from 15 to 43 min each minute.The injection volume was 100 mL.Totally 33 fractions were collected and concentrated until dryness using a rotary evaporator at 40 C.

HPLC-DAD-ESI-HRMS analyses of extracts and collected fractions
Crude extracts and collected and concentrated fractions were subjected to chromatographic analysis on a Dionex UltiMate 3000 HPLC system (Thermo Scientific, Sunnyvale, CA, USA) coupled with a diode array detector and high-resolution mass spectrometer with electrospray ionization source (ESI-HRMS; Impact HD Mass Spectrometer, Bruker, Billerica, MA, USA).The separations were performed on a reversephase column (Phenomenex Kinetex C18 column, 150 Â 4.6 mm, 2.6 lm) as described above in the section Qualitative and quantitative analyses of pigments.The source parameters were as follows: the spray needle voltage was set at 4.2 kV, nitrogen was used both as the nebulizing gas (3 bar) and the drying gas (12 L/min) and the temperature was 210 C. The scanning range was 50-2,000 m/z operating in the positive ion mode and the UV-Vis spectra were recorded from 200 to 700 nm.

Direct derivatization of fatty acids and quantitative analysis of fatty acid methyl esters by GC-FID
Methyl esters of fatty acids (FAME) were prepared as follows.Twelve selected fractions were resuspended in 80% acetone.One-fifth of the sample was withdrawn to a screwcap glass test tube.In total, 50 lg of internal standard (glycerol-tripentadecanoate, Sigma Aldrich, USA) was added and left to evaporate.Then 1 mL of 3 M HCl-Methanol (Sigma-Aldrich, USA) was added.Samples were sonicated for 10 min to achieve homogenous suspension, then transferred into preheated heat-block (Major Science, USA) and let stand at 90 C for 90 min.Samples were withdrawn from the heating-block, allowed 10 min to cool down to room temperature, and 2 mL of n-hexane were added.Samples were vortexed, sonicated for 10 min, and 2 mL of ice-cold 1 M NaCl (P-lab, Czech Republic) were added.Samples were vortexed again and centrifuged at 900 Â g for 10 min at 4 C.The upper n-hexane phase was carefully withdrawn and transferred into a crimp-top vial.Samples were directly transferred to HTA autosampler and analysis of fatty acid methyl esters via GC-FID was performed.
Quantitative and qualitative analysis of the FAME complements were performed by using a GC-FID (Trace 1300, Thermo) equipped with a flame ionization detector (FID) and connected to an autosampler (HTA, Italy).A TR-FAME column (60 m Â 0.32 mm, df 0.25 lm) was used.Helium was used as a carrier gas, at a constant flow of 2 mL/min.The temperature ramp was the following: the starting temperature was 140 C; it was increased to 240 C at the rate of 4.5 C/min and then maintained at 240 C for 10 min.The injector was kept at 260 C and the detector at 250 C. The retention times of FAMEs were compared to known standards (SupelcoV R 37 Component FAME Mix; PUFA No.3 Supelcofrom menhaden oil).The amounts of individual fatty acids were calculated using internal standards with a known heptadecanoic acid (C17:0) (Sigma Aldrich, USA) content and corrected by multiplying the integrated peak areas by the correction factors of the FID response.

P. larvae cultivation
P. larvae strains (CCM 4483, CCM 4484, CCM4485, CCM 4486, CCM 4487, CCM 4488, CCM 5680) were obtained from the Czech Collection of Microorganisms (CCM, Masaryk University, Brno, Czech Republic) and stored at À80 C. In addition, two field P. larvae isolates from the Czech Republic were used for testing (2018/1, 2018/2).Bacterial strains were cultured according to a previously published procedure (de Graaf et al., 2015) on MYPGP agar routinely used for P. larvae cultivation.Cultivation of vegetative P. larvae cells was performed under aerobic conditions without CO 2 for 48 h at 37 C in the dark.

Genotyping of P. larvae strains
All P. larvae strains were tested to determine individual ERIC genotypes as described (Genersch et al., 2006).Extraction of bacterial DNA of the P. larvae isolates and PCR reaction based on the 16S rRNA gene were performed with small modifications according to the previously described protocol (Bassi et al., 2015).ERIC I and ERIC II genotypes were distinguished according to the reported specifications Genersch et al. (2006), for details see Supplementary Material S2.

Determination of antimicrobial activity by spoton lawn antimicrobial assay
Suspension of freshly grown bacteria at a concentration of 0.5 McFarland was used for antimicrobial assays.The used concentration of 0.5 McFarland corresponded with 1-5 Â 10 7 CFU/mL varied according to the bacterial strain used.The dry mass of water, acetone, or methanol extracts were diluted to the concentration of 10 mg/mL with water, acetone, or methanol, respectively.To test water extracts, suspension of P. larvae was inoculated with a cotton wool stick onto a surface of MYPGP agar plate.For antimicrobial tests with methanol and acetone extracts MYPGP agar was heated to 42 C and mixed with 2 mL of bacterial suspension, then 50 mL of the mixture was poured onto Petri dishes.The volume of 10 mL of particular extract at the concentration of 10 mg/mL was applied directly onto Petri dishes.Control tests were performed with water, acetone, or methanol according to the type of extract used.The same volume as used for extracts was used when control spots were applied.Agar plates were incubated for 48 h at 37 C in the dark.Inhibition of P. larvae strains by tested extracts was evaluated by the diameter of the clear zone without P. larvae growth.All experiments were done in triplicate with two repetitions for each P. larvae strain tested.For evaluation of antimicrobial effect, diameters of inhibition zones were averaged for each ERIC genotype.

Antimicrobial test on P. larvae spores
For the observation of the effect of tested extracts on the spore form of P. larvae, preliminary tests on spores were done.The spore suspension was isolated from the hive debris of infected honey bees by the Tween 80 method according to (Bzdil, 2007).Spore concentration was determined by cultivating serial dilutions.Suspension of isolated spores was used for antimicrobial testing with a concentration of 1.2 Â 10 6 CFU/mL.Antimicrobial tests were performed as described above, 200 lL of spore suspension was used for inoculation of agar plates, and 10 mL of extracts at the concentration of 10 mg/mL was applied directly onto Petri dishes.Antimicrobial testing was done in triplicates for every extract.

Determination of minimum inhibitory concentration (MIC)
The MIC of selected C. vulgaris fractions separated by RP-HPLC was determined by the broth microdilution method (Eloff, 2019;Wiegand et al., 2008).The dry mass of extracts was dissolved in 80% acetone to a concentration 1000 mg/mL.Stock solutions at concentration 1000 mg/mL were prepared in 80% acetone and then diluted with MYPGP medium in ratio 60:20 (v/v).This solution was two-fold diluted into microtubes with MYPGP medium to obtain eight concentrations (5.9-750 mg/mL).Then, 10 mL of each concentration was transferred to a 96-well microplate (TPP, Switzerland) to obtain a final concentration range of 0.4-50 mg/mL for Chlorella CP extract and 0.03-4% of acetone.Bacterial cells were suspended in MYPGP medium to optical density OD 600 ¼ 0.1.Each well contained 140 mL of bacterial suspension and 10 mL of Chlorella CP extract.The negative growth control (150 mL MYPGP medium) and positive growth control (150 mL bacterial suspension) were included in each microplate.The effect of acetone on the viability of bacterial cells was controlled by adding 4% acetone to viable cells (144 mL bacterial suspension þ 6 mL 100% acetone).Tetracycline was used as an antibiotic positive control at a final concentration 66 mM.Microplates were incubated for 16 h at 37 C. Optical density was recorded at 600 nm by Synergy H1 reader (BioTek, Germany) every 15 min.The MIC was determined as the lowest concentration showing no bacterial growth.

Relative cell viability assay with resazurin
Cell viability after MIC determination in microplate was confirmed by a resazurin assay (Mann & Markham, 1998;Mariscal et al., 2009).Resazurin salt (Sigma Aldrich, USA) at concentration 0.15 mg/mL was added to each well (16 mL).Then the microplate was incubated at 37 C for 40 min in dark.The volume of each microplate well was transferred into a microtube and centrifuged 10 min at 14 000 Â g at laboratory temperature.The supernatant was used for the determination of cell viability by measuring fluorescence at Synergy H1 reader, excitation 560 nm, emission 590 nm).The viability was normalized to a positive control (100% viability).

C. vulgaris exposure to Apis mellifera larvae
Effect of the Chlorella microalgae on mortality of the young bee larvae was tested in vitro on reared honey bee larvae.Laboratory exposure bioassays were performed as described in Crailsheim et al. (2015).Briefly, first instar honey bee larvae were grafted into wells of 96-well-plates containing standard artificial larval diet (control groups) or larval diet containing C. vulgaris.Larvae were fed every day with the given solutions for 6 days.Specifically, diet for control group consisted from feeding solution A (6% w/v D-glucose, 6% w/v D-fructose, 1% w/ v yeast extract, 50% w/v royal jelly; Carl Roth, Germany and V cel ı produkty Karel Kol ınek company as a producer of royal jelly, Czech Republic) for day 1-2; solution B (7.5% w/v D-glucose, 7.5% w/v Dfructose, 1.5% w/v yeast extract, 50% w/v royal jelly) for day 3; and solution C (9% w/v D-glucose, 9% w/v D-fructose, 2% w/v yeast extract, 50% w/v royal jelly) for day 4-6 (Aupinel et al., 2005).Experimental groups received the same feeding solutions with the addition of Chlorella extract (1% w/v powder of Chlorella acetone or methanol extract in the diet solution).Dead larvae were counted every day of the experiment, the assay was performed in biological duplicate.

Statistical analysis
The normality was tested by the Kolmogorov-Smirnov test.Data were log 10 transformed because of the non-normal distribution of inhibition zones.Statistical analysis was evaluated in OriginPro 9 by 3way ANOVA and Tukey post hoc test.Survival analysis was calculated in OriginPro 9 by Kaplan-Meier estimation.All analyses were performed at a 0.05 level of significance.Figure 3 was prepared by using Flourish Studio (https://flourish.studio/).

Results
In this study, we tested the antimicrobial activity of C. vulgaris extracts concerning the Chlorella cultivation system by using different extraction solvents.Preparative and analytical analysis of the most active C. vulgaris extract provided a list of candidate compounds associated with the antimicrobial properties against P. larvae.

Genotyping of P. larvae strains
We genotyped the P. larvae strains used in the experiments for consistency with generally used nomenclature (Genersch et al., 2006).We analysed the commercially available strains, and six P. larvae strains were found to belong to the ERIC I genotype, whereas one strain belonged to the ERIC II genotype.Two field isolates belonged to the ERIC II genotype (strains 2018/1 and 2018/2) (Supplementary Material S2).

Antimicrobial activity of C. vulgaris extracts concerning chlorella cultivation and extraction methods
The C. vulgaris extracts showed varying antimicrobial activity against P. larvae depending on the extraction solvent and cultivation system used for algae (3-way ANOVA, Figure 1, Table 1).The highest activity was recorded for acetone extracts (diameter of inhibition zone ¼ 23.8 ± 7.5 mm, n ¼ 117) and methanol extracts (diameter of inhibition zone ¼ 22.0 ± 7.1 mm, n ¼ 130).Water extracts showed the lowest inhibition activity (diameter of inhibition zone ¼ 8.1 ± 0.9 mm, n ¼ 141) (Figure 1, Table 1).Regarding the cultivation system for alga biomass, the use of the CP and CPD cultivation systems resulted in an approximately 2Â higher antimicrobial activity compared to that of CH biomass in acetone and methanol extracts.From our results, P. larvae strains of the ERIC I genotype were more sensitive to Chlorella extracts compared to those of the ERIC II genotype (overall comparison based on 3-way ANOVA, Table 1).

Effect of C. vulgaris extracts on P. larvae spores
When we tested the inhibitory effects of the extracts on P. larvae spores, antimicrobial activity similar to that for vegetative P. larvae cells was observed.The application of CP and CPD acetone extracts resulted in the highest inhibition, and the application of water extracts resulted in the lowest inhibition (Table 2).

Fractionation of CP acetone extract
As the CP acetone extract contained the highest total concentration of pigments compared to those of the other two extracts (see Supplementary Data S1 for methods and results of pigment analysis) and additionally showed high antimicrobial activity, further fractionation of this extract was performed.Fractionation of the CP extract provided 33 fractions, which were further analysed for determining their antimicrobial activity.The HPLC-DAD chromatogram (recorded at 440 nm) is shown in Figure S2 (Supplementary Material S3).

Antimicrobial activity and minimum inhibitory concentrations (MICs) of the chromatography fractions
The antimicrobial activity of 33 fractions of the CP acetone extract was tested by performing spot-onlawn antimicrobial assays.Altogether, 9 fractions from the total 33 fractions showing retention times spanning 14-20 min and 24, 26, and 27 min produced detectable inhibition zones against strains of both P. larvae genotypes (fraction numbers 5-10, 14, 16, and 17) (Figure 2A).
For selected fractions of the CP acetone extract, we further determined the MIC and viability of P. larvae cells in tested samples using resazurin assays (Figure 2B).Fractions with low MIC values (fractions number 5-10) were associated with a low percentage of viable bacterial cells in resazurin assays.The lowest MIC value was detected for fraction number 7 (6.3 mg/mL for strains of both genotypes).Fractions 16 and 17 had the highest MIC values (Figure 2B).However, the viability of the P. larvae cells tested using fraction number 17 did not fully correspond to its high MIC value.

Identification of candidate antimicrobial compounds in fractions of CP acetone extract
As the antimicrobial activity was shown by 9 fractions of the CP acetone extract, further analytical characterization of these fractions was performed.HPLC-DAD-ESI-HRMS analysis identified the candidate bioactive compounds present in fractions 5-10.The main constituent of fraction 7 according to the MS signal was the compound with an m/z of 353.2686 (calculated for C 21 H 37 O 4 þ ).This compound was also found in the active fractions 5-8 ) in fractions 9 and 10, respectively.Based on molecular formulae, these compounds may be identified as monoacyl glycerols, and free fatty acids or methyl esters of fatty acids (Table 3, Figure 3).The presence of bound or free fatty acids in active fractions was determined via GC-FID analysis with prior methyl esterification.Fraction 7 contained linolenic acid (C18:3n3) as the major representative of this chemical group (42% of 13 fatty acids determined), and this fatty acid was also present in fractions 5, 6, and 8 (Figure 4).When additionally considering the molecular formula obtained using HRMS, this parent compound was identified as monolinolenin, a glycerol derivative of linolenic acid.The glyceroyl moiety of monolinolenin is removed during the  esterification procedure enabling detection of the methyl ester of the free fatty acid.A similar finding was observed for fractions 8 and 9.The molecular formula of the compound obtained using HRMS (C 21 H 39 O 4 þ ) suggested a presence of a monoacyl glycerol derivative containing two double bonds on the fatty acid chain.GC-FID analysis confirmed the presence of C18:2n6.However, quantitative GC-FID analysis revealed that this fatty acid was only a minor component in these fractions (approximately 3%), and the major fatty acids were identified as C16:2 in fraction 8 (34.25%) and C16:0 in fraction 9 (27.31%).These fatty acids were retrospectively identified via HPLC-HRMS analyses of these fractions in a free form and bound to methyl esters.Since there was an overlap in the mass of sodium adduct of palmitic acid (C16:0) and the protonated molecule of the methyl ester of linolenic acid (m/z 293.2475 vs. 293.2451),as well as an overlap in their retention times, these compounds could not be distinguished  ).These fractions showed smaller inhibition zones in antimicrobial assays than those of the above-mentioned fractions.Due to the complexity of fractions 16 and 17, no particular candidates were suggested.However, GC-FID analysis revealed the presence of several fatty acids (mainly C16:0 and C18:3n3).

Administration of C. vulgaris as a feed to A. mellifera larvae
As C. vulgaris is known to be consumed by honey bees and Chlorella extracts showed antimicrobial activity against P. larvae, we tested its application directly by administering it as a feed to bee larvae.The effect of C. vulgaris added into the larval diet on larval development and survival was tested in vitro.
Incorporation of powdered 1% acetone and 1% methanol extracts of C. vulgaris in the larval diet resulted in survival rates of 61.4% and 51.0%, respectively, after 6 days of feeding, compared to the 92.7% survival rate of the control group (Figure 4).

Discussion
Previous studies have reported the positive effects of feeding Chlorella spp.biomass to honey bees (Jehl ık et al., 2019;Ricigliano, 2020).Based on these findings, Chlorella spp.may be recommended as a potential pollen supplement (Eremia et al., 2013).Possible effects of C. vulgaris extracts against a P. larvae bacterial pathogen model were analysed in this study.As extracts inhibited the growth of P. larvae, active extracts were further analysed.This study identified candidate compounds associated with the observed antimicrobial activity.

Cultivation of C. vulgaris
Various cultivation systems have been optimized for the stable production of C.vulgaris at an industrial scale which is independent of weather or other climatic conditions (Doucha & L ıvansk y, 2011;Masojidek & Prasil, 2010).The cultivation methods used for Chlorella spp.are known to affect the pigment composition and content in the resulting biomass (Hu et al., 2018).While the biomass of phototrophically grown Chlorella (CP) is dark green, the biomass of heterotrophically grown Chlorella (CH) is bright green to yellowish colour due to the absence of chlorophylls.When CP is exposed to light radiation, the chlorophylls are degraded into pheophorbides and the biomass turns grey (CPD).Here, we used two types of C. vulgaris biomass cultivation systems because we expected that variability in pigment production could result in varying antimicrobial activity (Hu et al., 2018).Our results showed that CP and CPD extracts were associated with a higher production of pigments compared to that of extracts from the CH cultivation system.It is generally known that heterotrophic growth is associated with a limited ability to produce light-induced metabolites such as photosynthetic pigments (Perez-Garcia et al., 2011).This is in agreement with our results.CP and CPD cultivation resulted in two-fold higher antimicrobial activity compared to that obtained via CH cultivation, which may be associated with a higher pigment content.There is a limited number of studies focusing on the antimicrobial properties of algal pigments (Silva et al., 2020); however, the antimicrobial activities of individual pigments such as fucoxanthin have been reported previously (Karpi nski & Adamczak, 2019).

Extraction of potentially active compounds from C. vulgaris biomass
Three extraction strategies were tested to obtain extracts containing various hydrophilic or hydrophobic compounds for further testing of antimicrobial activity.Similar to the study by Cowan (1999), we used methanol, acetone, and water for the extraction of Chlorella biomass.The use of 70% methanol for biomass extraction was selected based on previous trials, as well as its demonstrated capacity to efficiently extract various bioactive compounds from cyanobacteria (Hrouzek et al., 2012;Kapu scik et al., 2013;Tomek et al., 2015;Vor a cov a et al., 2017).As acetone dissolves various hydrophobic and lipophilic components, acetone extraction is commonly used for the extraction of pigments and other non-polar compounds.Acetone has been previously described as the most efficient extraction solvent for plant antimicrobial components (Eloff, 1998) and is also considered a promising solvent for green microalgal extracts (Jebasingh et al., 2011;Shannon & Abu-Ghannam, 2016).Finally, water-based extraction was used as a negative control due to its low performance in the extraction of amphiphilic and lipophilic compounds (Cowan, 1999).As expected, the water extracts exhibited the lowest antimicrobial activity towards P. larvae.
Antimicrobial activity of C. vulgaris extracts against P. larvae Antimicrobial activity against P. larvae vegetative cells was demonstrated by all C. vulgaris extracts tested in this study.For vegetative cells, the extracts displayed a bactericidal effect, as there was no growth of P. larvae after its re-cultivation from inhibition zones (data not shown).We determined the antimicrobial activity towards the two most common P. larvae genotypes, ERIC I and ERIC II (Loncaric et al., 2009).In our study, different sensitivities of the ERIC genotypes were observed; however, the molecular background associated with this difference remains unclear.Additionally, we tested the antimicrobial activity via spot-on-lawn assays against P. larvae spores, which mimics the situation in the hive where honey bee larvae are infected by spores and not vegetative cells of P. larvae (de Graaf et al., 2015).Young bee larvae are highly susceptible to bacterial infections, and only a few P. larvae spores are sufficient to cause a disease outbreak (Brødsgaard et al., 1998).In contrast, older larvae are better protected by their developed immune defences (Chan et al., 2009).Our results showed that C. vulgaris extracts possessed anti-germination activity against P. larvae spores.A delay in spore germination can increase the survival rate of larvae at later stages of their development.Our results support the idea of using C. vulgaris as a prophylactic food supplement in beekeeping; however, clinical trials using bee colonies are required for further verification.

Antimicrobial activity of fractionated CP acetone extract
Based on the qualitative HPLC-DAD analysis of the extracts, we identified differences in the chromatographical patterns based on extraction solvents but not different cultivation systems (CP, CPD, and CH).
A further investigation of CP acetone extract involved the screening of the antimicrobial activity of 33 chromatographic fractions.Antimicrobial analyses resulted in the selection of 9 fractions with antimicrobial activity, which were further subjected to chemical analysis via HPLC-DAD-HRMS.Fraction number 7 was the most promising candidate for further analysis.Spot-on-lawn assays demonstrated the highest activity associated with this fraction against both P. larvae genotypes (30 mm inhibition zone), which was also confirmed via broth dilution methods (MIC 6.3 mg/mL).This MIC value is much higher compared to those of conventional antibiotics such as tylosin, which shows a MIC in the range of 0.0078-0.5 mg/mL for P. larvae (Alippi et al., 2005).The antimicrobial activity of C. vulgaris extracts is likely associated with the fatty acid content and its derivatives.Therefore, there is a limited risk of developing harmful residues in honey bee products.Moreover, the risk of P. larvae developing resistance to C. vulgaris extracts is potentially lower than the risk of developing resistance to commercially available antibiotics (Reynaldi et al., 2010).

Fatty acids and their derivatives as candidate compounds associated with the antimicrobial activity of CP acetone extract
Mono-acyl glycerols or free fatty acids were identified according to the molecular formulae of the predominant compounds detected via massspectrometric analysis (HPLC-DAD-HRMS) of the most active fractions.The antimicrobial activity of fatty acids, including linoleic and linolenic acid, against P. larvae has been previously shown by Feldlaufer et al. (1993).The most active fractions comprise the candidate compounds monolinolenin and monolinolein.Free linoleic and linolenic acids show antimicrobial activity, as reported by Catarina Guedes et al. (2011).
Furthermore, both marine and freshwater algae possess a high content of phenolic compounds showing numerous biological activities that are applicable in anticancer, antioxidant, or antimicrobial treatments (Li et al., 2011;Machu et al., 2015;Thomas & Kim, 2011).Phenolic compounds show antimicrobial activity against Gram-positive and Gram-negative bacteria (Escherichia coli, Pseudomonas aeruginosa, Staphylococcus aureus, Enterococcus faecalis, Vibrio fluvialis) when organic extracts of algae are applied (Al-Saif et al., 2014;Mohy El-Din & El-Ahwany, 2018).However, in this study, the chromatographic fractionation led to the separation of phenolic compounds from the tested fractions owing to their polarity.Therefore, we can assume that phenolic compounds were not responsible for the observed antibacterial effect against P. larvae.Catarina Guedes et al. (2011) reported longchain fatty acids as the key antimicrobial compounds present in cyanobacterial extracts.These findings, together with the absence of phenolic compounds in the analysed fractions of C. vulgaris extracts in our study, support the hypothesis that fatty acids may play an important role in the antimicrobial activity of C. vulgaris extracts against P. larvae.

Mortality of larvae supplemented with C. vulgaris extracts
The application of crude C. vulgaris has previously shown positive effects on honey bee physiology and colony development (Eremia et al., 2013;Jehl ık et al., 2019;Ricigliano & Simone-Finstrom, 2020); however, administering powder of C. vulgaris extracts to bee larvae in our in vitro experiment resulted in larval mortality.The high mortality rate in the experimental groups may be explained by the direct application of algae to larvae without previous digestion and chemical processing by honey bee workers, which occurs in bee hives under natural conditions.Therefore, a modified Chlorella-containing diet administered by adult worker bees may show different effects on larval fitness and mortality.

Conclusions
Previous studies have shown that the alga C. vulgaris has a positive impact on the health of both humans and animals, including honey bees.In this study, we confirmed the antimicrobial activity of C. vulgaris against the honey bee pathogen P. larvae, supporting the potential of Chlorella algae as a natural pollen supplement.Phototropic cultivation of algal biomass followed by acetone extraction was selected as the most appropriate technology for the extraction of antimicrobial agents.Additionally, monolinolenin, methyl esters of linolenic and free fatty acid linoleic acid, and/or palmitoyl methyl ester were identified as candidate antimicrobial compounds.In summary, our findings show that C. vulgaris has not only nutritional value for honey bees but can also be potentially used for prophylactic treatment of honey bee diseases.The application of Chlorella-based products needs to be further tested and validated in controlled field trials with AFB infection.

Figure 1 .
Figure 1.(A) Diameters of the inhibition zones of strains of P. larvae produced by different solvent extracts of CP, CH, and CPD.Strains of P. larvae were classified as EI or EII genotypes.CP, Chlorella biomass grown in phototropic conditions; CH, Chlorella biomass grown in heterotrophic conditions; CPD, degraded Chlorella biomass grown in phototropic conditions; EI, ERIC I genotype; EII, ERIC II genotype.

Figure 2 .
Figure 2. (A) Diameters of the inhibition zones detected for 33 fractions obtained via chromatographic separation of acetone CP extract for P. larvae strains of ERIC I and ERIC II genotypes.(B) MIC determination and relative bacterial cell viability (%) for highly active fractions.Fraction nr, fraction number; CP, Chlorella biomass grown in phototropic conditions; MIC, minimum inhibitory concentration.

Figure 3 .
Figure3.The proportion of 13 fatty acids in the total amount of fatty acids in fractions of CP Chlorella acetone extract as determined via GC-FID (Supplementary Data S4).CP, Chlorella biomass grown in phototropic conditions; GC-FID, gas chromatography coupled with a flame ionization detector.

Figure 4 .
Figure 4. Honey bee larvae survival after exposure to algal extracts (acetone, methanol; 1%) incorporated in larval diet as estimated using Kaplan-Meier analysis.

Table 1 .
The overall results from a 3-way ANOVA test performed using log 10 transformed data.

Table 2 .
Diameters of inhibitory zones produced by the application of Chlorella extracts on P. larvae spores.¼ the methanol CPD extract was expended in the experiments on P. larvae vegetative cells; therefore, it was not possible to obtain a new identical CPD methanol extract from the same batch.Preparing a new batch of CPD methanol extract would require a new batch of Chlorella biomass, which was not available at the time of our study.

Table 3 .
Predicted antimicrobial candidates in fractions of CP Chlorella acetone extract.
via HPLC-DAD-HRMS analysis.C18:2n6 was found to be the major fatty acid component in fraction 10 (44.79%); the compound was identified as a methyl ester of linoleic acid.