The Fly Factor Phenomenon Is Mediated by Interkingdom Signaling Between Bacterial Symbionts and Their Blow Fly Hosts

Insect Science (2018) 00, 1–10, DOI 10.1111/1744-7917.12632

ORIGINAL ARTICLE The ﬂy factor phenomenon is mediated by interkingdom signaling between bacterial symbionts and their blow ﬂy hosts

Yonathan Uriel , Regine Gries, Lorna Tu, Cassandra Carroll, Huimin Zhai, Margo Moore and Gerhard Gries Department of Biological Sciences, Simon Fraser University, Burnaby, British Columbia, Canada

Abstract We tested the recent hypothesis that the “fly factor” phenomenon (food currently or previously fed on by flies attracts more flies than the same type of food kept inaccessible to flies) is mediated by bacterial symbionts deposited with feces or regur- gitated by feeding flies. We allowed laboratory-reared black blow flies, Phormia regina (Meigen), to feed and defecate on bacterial Luria–Bertani medium solidified with agar, and isolated seven morphologically distinct bacterial colonies. We identified these using matrix-assisted laser desorption/ionization mass spectrometry and sequencing of the 16S rRNA gene. In two-choice laboratory experiments, traps baited with cultures of Pro- teus mirabilis Hauser, Morganella morganii subsp. sibonii Jensen, or Serratia marcescens Bizio, captured significantly more flies than corresponding control jars baited with tryptic soy agar only. A mixture of seven bacterial strains as a trap bait was more attractive to flies than a single bacterial isolate (M. m. sibonii). In a field experiment, traps baited with agar cultures of P.mirabilis and M. m. sibonii in combination captured significantly more flies than traps baited with either bacterial isolate alone or the agar control. As evident by gas chromatography-mass spectrometry, the odor profiles of bacterial isolates differ, which may explain the additive effect of bacteria to the attractiveness of bacterial trap baits. As “generalist bacteria,” P. mirabilis and M. m. sibonii growing on animal protein (beef liver) or plant protein (tofu) are similarly effective in attracting flies. Bacteria-derived airborne semiochemicals appear to mediate foraging by flies and to inform their feeding and oviposition decisions. Key words blow flies; enteric bacteria; fly factor; interkingdom communication; microbial symbionts; semiochemical attractants

Introduction feeding activity by flies promotes the recruitment of other flies to a food source (Danchin & Wagner, 1997; Lihoreau The “fly factor” refers to observations that food currently & Rivault, 2011; Holl & Gries, 2018). or previously fed on by flies attracts more flies than the The fly factor has been observed in diverse dipteran same type of food kept inaccessible to flies (Barnhart taxa, including black blow flies (Phormia regina Meigen) & Chadwick, 1953). Several studies have reported that (Dethier, 1955), house flies (Musca domestica Linnaeus) (Barnhart & Chadwick, 1953; Acree et al., 1959; Holl & Gries, 2018), face flies (Musca autumnalis De Geer) Correspondence: Yonathan Uriel, Department of Biological (Teskey, 1969), and green bottle flies (Lucilia sericata Sciences, Simon Fraser University, 8888 University Drive, Burn- Meigen) (Brodie et al., 2015). The fly factor is also ef- aby, British Columbia V5A 1S6, Canada. Tel: +1 778 782 5939; fective across taxa; for example, feeding black blow flies fax: +1 778 782 3496; email: [email protected] attract foraging house flies, and vice versa (Dethier, 1955),

1 C 2018 Institute of Zoology, Chinese Academy of Sciences 2 Y.Urieletal. and feeding green bottle flies attract foraging black blow Working with the black blow fly, Phormia regina,asa flies, and vice versa (Brodie et al., 2015). model species we tested the hypotheses that (i) specific, Previously, in investigating the mechanisms underly- fly-deposited bacteria are attractive to flies; (ii) multiple ing the fly factor we showed that neither temperature strains of fly-deposited bacteria are more attractive to flies nor relative humidity differed between feeding and non- than single-strain bacteria; and (iii) fly-attractive bacteria feeding flies. In contrast, feeding flies produced signif- are attractive to flies irrespective of the substrate the bac- icantly more CO2 than nonfeeding flies (Holl & Gries, teria grow on. 2018), suggesting that elevated levels of CO2 may sig- nify fly feeding activity and thus the presence of a food source or oviposition site. However, experimental test- Materials and methods ing of CO2 as a foraging cue for flies did not reveal any behavior-modifying effect (Holl & Gries, 2018), confirm- Rearing of experimental flies ing that house flies do not respond to CO2 (Richards, 1922; Wieting & Hoskins, 1939). Attraction of house flies to fed- We maintained blow fly colonies in the Simon Fraser on food and to house fly feces and regurgitate prompted University (SFU) insectary (Burnaby, BC, Canada), ° Holl and Gries (2018) to hypothesize that house flies re- at 25 C, 65% relative humidity, and a photoperiod of spond to minute amounts of semiochemicals (message- L14 : D10. We reared fly maggots on bovine liver bearing chemicals) produced by microbes in these sub- (Supreme Meat Supplies Ltd., Burnaby, BC, Canada) and strates. Their hypothesis was inspired by findings that provisioned adult flies with sugar, and, 3 d post-eclosion, feeding and defecating flies can inoculate food sources with milk powder ad libitum as a protein source. with symbiotic microbes (Hendrichs et al., 1992) present in their digestive tract (Gupta et al., 2012) and salivary Isolating and culturing of bacteria associated with flies glands (Singh et al., 2015). One of these microbes is the bacterium Proteus mirabilis Hauser (Gupta et al., 2012) To isolate bacteria associated with P.regina, we released that produces semiochemicals such as indole and dimethyl 50–60 flies into a metal mesh cage (61 cm3; BioquipR , trisulfide (Ma et al., 2012; Tomberlin et al., 2012), which Compton, CA, USA), which contained a water wick and are well-documented fly attractants (Mulla et al., 1977; Luria–Burtani agar (LBA) in an open Petri dish (85 mm × Cosse´ & Baker, 1996; Brodie et al., 2014, 2016). 15 mm). After flies had fed and defecated on the agar dish Proteus mirabilis colonizes animal feces and car- for1hat25°C, we removed and incubated it aerobically at rion that attract foraging flies (Chaudhury et al., 2010; 25 °C for 24 h. Thereafter, we restreaked morphologically Tomberlin et al., 2017). Interestingly, some fly-attractive distinct colonies onto new LBA plates, until we obtained odorants produced by P. mirabilis also serve as signaling pure colonies, and incubated them at 25 °C for 24 h. Plates molecules in its quorum sensing (QS) behavior (Sturgill were stored at 4 °C. We were able to isolate seven bacterial & Rather, 2004; Ma et al., 2012; Tomberlin et al., 2012), a colonies. We made storage cultures of each isolate and phenomenon whereby secondary metabolites secreted by stored them in 20% glycerol medium at –80 °C. the bacteria accumulate to a critical threshold concentration which then triggers transcription of swarming genes in bacterial conspecifics (Verstraeten et al., 2008). As QS Identification of bacteria is linked to bacterial proliferation (Kearns, 2010), the relative abundance of QS-associated odorants may signal the We identified five of the seven bacterial colonies to nutritional quality of a food source to foraging flies. the species level by mass spectrometry using matrix- Prior studies have not yet demonstrated a direct link be- assisted laser desorption/ionization (MALDI-TOF MS; tween the feeding activity of blow flies, their deposition of Bruker Corp., Billerica, MA, USA) (Ibarra Jimenez et al., microbial symbionts, and the attractiveness of these sym- 2018). We used a Bruker bacterial test standard (Bruker bionts to foraging flies. Furthermore, if microbes were to Corp., Millica, MA, USA) for calibration in accordance emit signature odorants, multiple strains of microbes in with manufacturer instructions. For each strain, we pro- combination (with a combined and more complex semio- cessed two preparations, and analyzed all spectra using chemical blend) might be more readily detectable and Biotyper software (Bruker Corp., Billerica, MA, USA). thus be more attractive to flies than single-strain bacteria. This Biotyper software calculates an arbitrary score for Finally, as generalist detritivores, we would expect blow each sample between 0 and 3 by comparing sample mass flies to respond to the odorants from generalist bacteria spectra to reference mass spectra; we accepted species growing on diverse detritus. assignments at scores of >2.0 in accordance with the

C 2018 Institute of Zoology, Chinese Academy of Sciences, 00, 1–10 The fly factor is microbe mediated 3 manufacturer’s recommended protocol. We identified the we incubated all slices of beef liver and tofu for 24 h at remaining two colonies to the genus level by amplify- 25 °C. ing and sequencing the 16S rRNA gene, and by comparing the resulting partial sequences to the database of the National Center for Biotechnology Information Gen- Specific experiments Bank, using the online BLASTn tool (Bethesda, MD; http://www.ncbi.nlm.nih.gov/BLAST.cgi). Hypothesis 1: Fly-deposited bacteria are attractive To prepare bacteria for sequencing, we cultured them in to conspecific flies Experiments 1–7 followed the general LB broth at 25 °C for 18–20 h. We extracted chromosomal laboratory two-choice experimental design (see above). In DNA using a standard phenol/chloroform extraction pro- each experimental replicate, we baited one of the paired tocol (Sambrook et al., 1989), and amplified 1.5 kb of the 450 mL amber glass jars with a 1/4 section cut from 16S rRNA gene by polymerase chain reaction using the a Petri dish (85 mm diameter) containing sterile TSA Eubacteria-specific primers 16F27 (5ʹ-CCA GAG TTT (control stimulus) and the other with a 1/4 section of GAT CMT GGC TCA G-3ʹ) and 16R1525XP (5ʹ-TTC TSA previously inoculated with a 24 h culture of one TGC AGT CTA GAA GGA GGT GWT CCA GCC-3ʹ), of the bacterial isolates, that is, Exiguobacterium spp. A following the PCR protocol from Pidiyar et al. (2004) with (Exp. 1; n = 20), Exiguobacterium spp. B (Exp. 2; n = 15), Kodaq DNA polymerase (Applied Biological Materials, Proteus mirabilis (strain A) (Exp. 3; n = 20), P.mirabilis Canada). (strain B) (Exp. 4; n = 20), Morganella morganii subsp. sibonii Jensen (strain A) (Exp. 5; n = 18), M. m. sibonii (strain B) (Exp. 6; n = 20), or Serratia marcescens Bizio General design for laboratory behavioral experiments (Exp. 7; n = 20). Hypothesis 2: Multiple strains of fly-deposited bac- We ran two-choice experiments 1–9 and 11–14 at 25 teria are more attractive to conspecific flies than a °C(± 1 °C) in the SFU insectary. For each experimen- single bacterial strain We tested hypothesis 2 in the lab- tal replicate, we placed two amber glass jars (450 or 900 oratory (Exps. 8–9) and in the field (Exp. 10). Parallel mL) 20 cm apart in a metal mesh bioassay cage (61 cm3; experiments 8–9 followed the general laboratory experi- BioquipR ) illuminated from above by fluorescent lights mental design described above. In each replicate of ex- (Phillips F32TA, Amsterdam, the Netherlands). The am- periment 8 and 9, we baited one of the paired 900 mL ber glass occluded visual cues of test stimuli inside jars, amber glass jars (larger jars were used to accommodate and a glass funnel placed on top of each jar allowed flies more agar) with a 1/7th section of each of seven TSA to enter but not exit the jar. By random assignment, we dishes each inoculated with one of the seven isolated bac- baited one jar in each pair with Tryptic Soy agar (TSA) terial strains (Exiguobacterium spp. A; Exiguobacterium (control stimulus; see detail below) and the other with spp. B; P.mirabilis [strain A]; P.mirabilis [strain B]; M. m. TSA that had been inoculated and aerobically incubated sibonii [strain A]; M. m. sibonii [strain B]; S. marcescens). (25 °C, 24 h) with an isolated bacterial strain (treatment The control stimulus in experiments 8 and 9 consisted stimulus). For each experimental replicate, we released a of an amber glass jar baited with one (Exp. 8) or seven mixed group of 50–60, 3- to 7-d-old male and female flies (Exp. 9) 1/7th sections of a TSA dish inoculated with M. into a cage. After 1 h, we placed bioassay cages in a –15 m. morganii (strain A). The contrasting control stimuli °C freezer to cold-euthanize bioassay flies, and counted in experiments 8 and 9 allowed us to distinguish whether all male and female flies captured within amber jars flies are truly more attracted to multiple strains of bacte- (responding flies) and still within cages (nonresponding ria, or whether attraction is based solely on the volume of flies). attractive bacteria. We chose M. m. sibonii (strain A) as Alternative control stimuli in experiments 11–14 con- the single bacterial strain control stimulus because of its sisted of small slices (30 mm × 30 mm × 10 mm) of demonstrated ability to attract flies (see Results). either beef liver (Supreme Meat Supplies Ltd, Burnaby, We ran experiment 10 at a farm in Abbotsford, BC BC, Canada) (Exps. 11–12) or tofu (Soyganic Super Firm between 21 and 22 August 2017, using a randomized Tofu, Sunrise Soya Foods, Vancouver, BC, Canada) (Exps. complete block design. We deployed custom delta traps 13–14) that we sanitized by rinsing them with 70% EtOH (100 mm wide × 87 mm high × 150 mm long) that we followed by autoclaved water. Corresponding treatment built from black poster board (Royal Brites Recycled stimuli consisted of sanitized slices of beef liver or tofu Black Poster Board, Staples) coating the inside surface that had been inoculated by streaking single cultures of an with adhesive (Tanglefoot; Tanglefoot Acquisition, isolated bacterial strain. Prior to the start of the bioassay, Grand Rapids, MI, USA). Within each block, we placed

C 2018 Institute of Zoology, Chinese Academy of Sciences, 00, 1–10 4 Y.Urieletal. four delta traps on the ground along the wall of buildings, camphor; 1-methoxy-4-methylbenzene [Sigma-Aldrich with intertrap and interblock spacing of 10 m. Each block Co., St. Lois, MO 63013, USA]; 2-phenylethanol [Fluka consisted of four traps, each of which received one of Chemie GmbH, CH-9471 Buchs, Switzerland]; benzyl al- four treatments by random assignment. These treatments cohol [Fisher Scientific, Fair Lawn, NJ 07410, USA]; N- consisted of a Petri dish (85 mm × 15 mm) that contained (3-methylbutylidene)-2-methyl-1-propylamine and N-(3- (1) TSA only, (2) TSA inoculated with P. mirabilis,(3) methylbutylidene)-3-methyl-1-butylamine [synthesized TSA inoculated with M. m. sibonii, or (4) half a TSA plate with reference to a previous report; Largeron et al., 2008]; inoculated with P.mirabilis and TSA (half a plate) inocu- N-(2-phenylethylidene)-3-methyl-1-butylamine [synthe- lated with M. m. sibonii. For this experiment, we cultured sized with reference to a previous report [Yamaguchi & both P.mirabilis and M. m. sibonii as described above. Takeda, 1992]). To test for the production of H2S, we inoc- Hypothesis 3: Fly-attractive bacteria remain attrac- ulated bacteria on triple sugar iron slants and observed the tive to flies irrespective of the type of protein-based results after 24 h of growth at 25 °C (MacFaddin, 2000). substrate they grow on Following the general laboratory two-choice experimental design, we ran two sets of parallel experiments. In each of parallel experiments 11 and Statistical analysis 12, we baited the control amber glass jar (450 mL) with sanitized beef liver and the treatment jar with sanitized We analyzed all data using R for Windows (R Core beef liver inoculated with either P. mirabilis (strain A) Team, 2017). For each experiment except 10 (see below), (Exp. 11) or M. m. sibonii (strain A) (Exp. 12). Similarly, we logit-transformed the raw proportions of flies captured in parallel experiments 13 and 14, we baited the control in treatment and control jars at experiment termination and jar with sanitized tofu and the treatment jar with sani- analyzed data by a Welch t-test. For Exp. 10, we compared tized tofu inoculated with either P. mirabilis (strain A) mean captures per trap bait type using a single-factor (Exp. 13) or M. m. sibonii (strain A) (Exp. 14). CRD ANOVA followed by a Tukey’s HSD test (Package “emmeans” v1.1.2; Leuth et al., 2018). For experiments 11–14, we performed a logistic regression to compare the Capture and identification of headspace odorants mean number of trap captures between the beef liver and and gases tofu bait treatments using a binomial generalized linear model followed by Tukey’s HSD test. We placed 10 mm × 10 mm × 30 mm slices of TSA, beef liver, or tofu that we had inoculated and incubated Results with specific bacterial isolates for 24 h into a cylindrical R × Pyrex glass chamber (180 mm high 90 mm wide). Identification of bacterial isolates A vacuum pump drew charcoal-filtered air at 0.5 L/min for 2 h through the chamber and through a glass column Using Bruker Daltonik MALDI Biotyper software × (6 mm outer diameter 150 mm) containing 200 mg (Bruker Corp., Billerica, MA, USA), we identified one TM of Porapak-Q (50–80 mesh) adsorbent (Byrne et al., bacterial isolate as Serratia marcescens, two isolates as 1975). We desorbed bacteria- and medium-derived odor- Morganella morganii subsp. sibonii (A, B), and two iso- TM ants captured on Porapak-Q with 2 mL of a pentane : lates as Proteus mirabilis (A, B). Through DNA extraction μ diethyl ether mixture (1 : 1) and analyzed 2 L aliquots and sequencing, we identified two further isolates as Ex- by a Saturn 2000 Ion Trap GC-MS (Varian Inc., now iguobacterium spp. (A, B) (GenBank accession numbers Agilent Technologies Inc., Santa Clara, CA, USA) oper- CP018057.1 and JN852813.1), each with 99% identity to ated in full-scan electron impact mode and fitted with a the reference strain. DB-5 GC-MS column (30 m × 0.32 mm inner diameter; J&W Scientific, Folsom, CA, USA). We used helium as the carrier gas (350 mm/s) with the following tempera- Testing attractiveness of bacterial isolates to flies ture program: 50 °Cfor1min,10°C/min until 280 °C (held for 10 min), setting the injector port and ion trap to Hypothesis 1: Fly-deposited bacteria are attractive 250 °C and 260 °C, respectively. Weidentified compounds to conspecific flies Neither Exiguobacterium A (Exp. 1) by comparing their retention indices (van den Dool & nor Exiguobacterium B (Exp. 2) incubated on TSA for Kratz, 1963) and mass spectra with those reported in the 24 h attracted more flies than corresponding TSA controls literature and with those of authentic standards (benzalde- (Fig. 1, Exp. 1: t(24) = 1.8, P = 0.08; Exp. 2: t(38) = 1.9, P = hyde; dimethyl trisulfide; phenol; indole; 1,8-cineole; 0.06). In contrast, all other bacterial strains were attractive

C 2018 Institute of Zoology, Chinese Academy of Sciences, 00, 1–10 The fly factor is microbe mediated 5

Fig. 1 Mean (+ SE) proportion of male and female Phormia regina captured in two-choice laboratory experiments 1–7 in jars baited with either tryptic soy agar (TSA) or TSA inoculated with Exiguobacterium spp. (congener A or B), Proteus mirabilis (strain A or B), Morganella morganii subsp. sibonii (strain A or B), or Serratia marcescens. For each experiment, an asterisk indicates a significant preference for the respective test stimulus (Welch’s t-test, P < 0.05), n indicates the number of replicates tested, and numbers inside bars indicate the mean proportion of the 50–60 flies that were released for each replicate and within the 1 h experimental period captured in the two jars. to flies: both P. mirabilis A (Exp. 3) and P. mirabilis B (Exp. 4) grown on TSA were more attractive to flies than corresponding TSA controls (Fig. 1, Exp. 3: t(38) = 9.9, P < 0.001; Exp. 4: t(38) = 25, P < 0.001). Similarly, M. m. sibonii A (Exp. 5) and M. m. sibonii B (Exp. 6) grown on TSA were more attractive than corresponding TSA controls (Fig. 1, Exp. 5: t(38) = 8.8, P < 0.001; Exp. 6: t(28) = 9.5, P < 0.001), as was S. marcescens (Fig.1,Exp. + 7: t(38) = 2.8, P = 0.008). Fig. 2 Mean ( SE) proportion of male and female Phormia Hypothesis 2: Multiple strains of fly-deposited bac- regina captured in two-choice laboratory experiments 8–9 in teria are more attractive to conspecific flies than a jars baited with tryptic soy agar (TSA) inoculated with a 1/7 single bacterial strain Seven bacterial isolates presented section of each of seven TSA dishes, each inoculated with one of seven bacterial isolates (Serratia marcescens, Morganella together (each grown on one 1/7th section of a TSA dish) morganii subsp. sibonii [strain A or B], Proteus mirabilis [strain attracted more flies than did M. m. sibonii Aasasin- AorB],orExiguobacterium congeners [A, B]) or baited with gle isolate grown either on one 1/7th section of a TSA one (Exp. 8) or seven (Exp. 9) 1/7th sections of a TSA dish dish and presented together with six sterile 1/7th sections inoculated with M. m. sibonii. The gray section in the TSA dish (Fig. 2, Exp. 8: t(38) = –2.0, P = 0.04) or grown on each of experiment 8 indicates agar void of bacterial inoculum. For of seven 1/7th sections of a TSA dish (Fig. 2, Exp. 9: each experiment, the asterisk indicates a significant preference < t(38) = –4.0, P < 0.001). for the mixture of bacteria (Welch’s t-test, P 0.05), n indicates In the field study (Exp. 10), traps baited with both the number of replicates tested, and numbers inside bars indicate P. mirabilis and M. m. sibonii on TSA (half a Petri dish the mean proportion of 50–60 flies that were released for each each) captured significantly more flies than traps baited replicate and within the 1 h experimental period captured in the with either P. mirabilis or M. m. sibonii (whole TSA two jars. dish each), or baited with a TSA blank control (ANOVA F3,44 = 6.9, P < 0.001, Fig. 3). The number of flies recorded specimens of six genera (Fannia, Calliphora, captured by the latter three treatments were not signifi- Melanodexia, Lucilia, Musca, and Thricops) in three fam- cantly different from one another (Fig. 3). ilies of flies (Fannidae, Calliphoridae, Muscidae). Fly captures were taxonomically diverse. Discounting Hypothesis 3: Fly-attractive bacteria attract flies taxa that were represented by three or fewer captures, we irrespective of the substrate they grow on The sets

C 2018 Institute of Zoology, Chinese Academy of Sciences, 00, 1–10 6 Y.Urieletal.

Fig. 3 Mean (+ SE) number of flies (mainly in the genera Fannia, Calliphora, Melanodexia, Lucilia, Musca,andThricops) captured in adhesive-coated Delta traps (see Methods for detail) in a field study (Exp. 10) run on a farm in the Lower Mainland of British Columbia between 21 and 22 August 2017. Traps within each of 12 experimental replicates received one of four treatments, consisting of a Petri dish that contained (1) sterile tryptic soy agar (blank TSA), (2) TSA inoculated with Proteus mirabilis, (3) TSA inoculated with Morganella morganii subsp. sibonii, or (4) TSA (1/2 of a dish) inoculated with P. mirabilis and TSA (1/2 of a dish) inoculated with M. m. sibonii. Box plots show the median, lower, and upper quartiles and ± whiskers of trap captures; different letters indicate statistically different trap captures (one-way ANOVA followed by Tukey’s HSD test for pairwise comparisons of means; P < 0.05).

Fig. 4 Mean (+SE) proportion of male and female Phormia regina captured in two-choice laboratory experiments in jars (A) baited with either sanitized beef liver or beef liver inoculated with Proteus mirabilis (Exp. 11) or Morganella morganii subsp. sibonii (Exp. 12), and (B) baited with either tofu or tofu inoculated with P. m i r a b i l i s (Exp. 13) or M. m. sibonii (Exp. 14). For each experiment, an asterisk indicates a significant preference for the respective test stimulus (Welch’s t-test, P < 0.05), n indicates the number of replicates tested, and numbers inside bars indicate the mean proportion of the 50–60 flies that were released for each replicate and within the 1 h experimental period captured in the two jars. of traps baited with beef liver inoculated with either regression model, microbe-innoculated tofu was more ef- P. mirabilis (Exp. 11) or M. m. sibonii (Exp. 12) each fective at attracting flies than microbe-innoculated beef captured significantly more flies than corresponding liver between experiments 11 and 14 (Wald’s χ 2 = 4.63, control traps baited with sanitized beef liver (Fig. 4; residual df = 73, P = 0.03). However, we do not consider Exp. 11: t(38) = 5.2, P < 0.001; Exp. 12: t(37) = 5.7, the actual difference in odds of choosing one source over P < 0.001). Similarly, the sets of traps baited with another as biologically relevant; the odds of a fly choosing tofu inoculated with either P. mirabilis (Exp. 13) or M. microbe-innoculated tofu over sanitized tofu was 6.34 : morganii (Exp. 14) each captured significantly more 1, wheras the odds of a fly choosing microbe-innoculated flies than corresponding control traps baited with san- beef liver over microbe-innoculated tofu was 3.90 : 1, re- itized tofu (Fig. 4; Exp. 13: t(38) = 25, P < 0.001; sulting in an odds ratio for microbe-innoculated tofu over Exp. 14: t(33) = 2.7, P = 0.01). Based on our logistic microbe-innoculated beef liver of 1.6 : 1.

C 2018 Institute of Zoology, Chinese Academy of Sciences, 00, 1–10 The fly factor is microbe mediated 7

Fig. 5 Total ion chromatograms of headspace volatile extracts obtained from the bacteria Proteus mirabilis (A), Morganella morganii subsp. sibonii (B), Serratia marcescens (C), and Exiguobacterium spp. (D) grown on tryptic soy agar. Volatile odorants were identified as follows: 1 = benzaldehyde, 2 = dimethyl trisulfide, 3 = phenol, 4 = 2-phenylethanol, 5 = indole, 6 = N-(3-methylbutylidene)-2- methyl-1-propylamine, 7 = N-(3-methylbutylidene)-3-methyl-1-butylamine, 8 = N-(2-phenylethylidene)-3-methyl-1-butylamine, 9 = 1-methoxy-4-methylbenzene, 10 = benzyl alcohol, 11 = unknown, 12 = 1,8-cineole, 13 = camphor. Note that the presence, and ratio, of known fly attractants (dimethyl trisulfide, phenol, 2-phenylethanol, indole), and of other odorants with heretofore undetermined attractiveness to flies, differ among bacterial isolates.

Capture and identification of headspace odorants We established a direct link between the feeding ac- and gases tivity of blow flies and the deposition of their bacterial symbionts by (i) allowing blow flies to feed on sterile GC-MS analysis of the headspace volatiles from bac- growth media, (ii) incubating fed-on media aerobically terial cultures revealed a number of odorants, which are until bacterial colonies formed, (iii) isolating and purify- known blow fly attractants, most notably indole, dimethyl ing morphologically distinct bacterial colonies onto new trisulfide, phenol, and 2-phenyl ethanol (Fig. 5). Only P. dishes, and (iv) identifying bacterial isolates by MALDI- mirabilis (strains A and B) produced H2S, based on the TOF mass spectrometry or by sequencing the 16S rRNA appearance of TSI-agar slants after 24 h of incubation gene. We isolated bacteria from four genera: specifically, (MacFaddin, 2000). one strain of Serratia marcescens, two strains each of M. morganii subsp. sibonii and Proteus mirabilis, and two Exiguobacterium spp. It has been shown that the Discussion digestive tract of house flies harbors a diverse community of microbes including P. mirabilis and M. morganii Our data support the conclusion that the fly factor phe- (Gupta et al., 2012), hence our bacterial isolates most nomenon, in which food fed on by flies is more attractive likely originated from fly feces. However, it is possible to flies than untouched food, is mediated in part by bacte- that these bacteria may have originated from the salivary rial symbionts deposited by feeding flies on a food source gland of flies. Liquid salivary regurgitate disseminated as part of regurgitate and feces. As the bacteria prolifer- by the sponging proboscis of flies dissolves solid foods ate, they secrete secondary metabolites, including specific such as agar before it can be ingested. Therefore, any mi- odorants that attract foraging flies. Below, we elaborate crobes present in these regurgitates may inoculate food on this conclusion. being fed on by flies. This is supported by the finding that

C 2018 Institute of Zoology, Chinese Academy of Sciences, 00, 1–10 8 Y.Urieletal.

P.mirabilis is indeed present in the salivary gland of bottle sensing odorants emitted by P. mirabilis (Ma et al., fly maggots (Ma et al., 2012). 2012), most notably indole and putrescine, are metabo- When we collected bacteria from feeding blow flies, we lites from amino acid breakdown pathways not specific relied entirely on culture-based methodology, thereby in- to P.mirabilis (Lee & Lee, 2010; Schneider & Wendisch, evitably limiting the diversity of bacteria we could isolate 2011). This may explain why odorants from nonswarming and characterize; it has been estimated that only <1% of bacteria (M. m. morganii) were also capable of attracting all bacteria species can be isolated using culture-based flies (Fig. 1). These specific metabolites may also cor- methods (Staley & Konopka, 1985). Even though we relate to the breakdown of proteins within the substrate, could not capture the true complexity of the interactions which may make the substrate more palatable for flies. between blow flies and their diverse microbiota, we still The blow fly-deposited microorganisms are generalist provide conclusive evidence for interkingdom signaling bacteria, that is, they can grow on diverse substrates in- between bacteria and blow flies. In behavioral laboratory cluding animal carrion, feces, beef liver, and tofu. All of bioassays, individual bacterial isolates differed in their these substrates are rich in protein that bacteria can digest, ability to attract flies. Cultures of Serratia marcescens in the process producing volatile secondary metabolites and each of the two M. m. sibonii and P. mirabilis strains that signal foraging flies. For these signals to reliably con- attracted flies, whereas each of the two Exiguobacterium vey information about a resource, their quality or strength species exhibited no significant attractiveness to flies should not be affected by the origin of the protein that over a sterile agar control (Fig. 1). As expected, all is metabolized by bacteria. In our study, bacterial iso- seven bacterial isolates grown separately but presented in lates growing on sanitized beef liver (animal protein) or combination attracted more flies than did M. m. sibonii sanitized tofu (plant protein) were similarly effective at on its own, regardless of the concentration of M. m. si- attracting foraging flies (Fig. 4). Flies which then feed on bonii (Fig. 2). Preferential responses of flies to a mixture these substrates can potentially ingest bacteria and vector of bacterial strains over single isolates was also found in them to new resources (Moriya et al., 1999; Junqueira the field experiment (Fig. 3). As a trap bait, the combi- et al., 2017). nation of P.mirabilis and M. m. morganii resulted in sig- Bacterial odorants have been shown to serve as cues nificantly higher captures of flies than either of the two that prompt aggregation and oviposition behavior of many bacterial strains alone or the agar control treatment. Eddy dipterans, including mosquitoes (Ponnusamy et al., 2008), et al. (1975) obtained similar results with the screwworm various blow flies (Tomberlin et al., 2017), black soldier Cochliomyia hominivorax, where a mixture of bacteria flies, Hermetia illucens (Zheng et al., 2013), onion flies, (P. mirabilis, P. vulgaris, P. rettgeri, and M. morganii) Delia antiqua (Judd & Borden, 1992), and house flies attracted more screwworms than single bacterial strains. (Lam et al., 2007). For those flies that apparently lack en- The diversity of microbial communities on a food source dogenous long-range pheromones (Stoffolano et al., 1997; may reflect its nutritional complexity, with nutrient de- Brodie et al., 2015), bacterial odorants may substitute for ficient resources being less capable of sustaining diverse pheromone signaling to elicit aggregation or oviposition microbial communities. Alternatively, a diverse microbial behavior. community may be an indirect indicator of the extent to In summary, we show that semiochemicals produced which a food source has been colonized and consequently by fly-deposited bacteria are responsible for the fly factor enriched in microbial nutrients. phenomenon, and that combinations of bacterial strains Bacteria-derived airborne semiochemicals may consti- are more effective than single strains. These findings tute the most direct cues for resource localization by flies provide the basis for the development of commercial fly and inform their feeding and oviposition decisions. The baits comprising specific bacterial isolates and appropri- composition of microbial headspace volatiles would seem ate growing media. to reflect the diversity of microbial growth on a resource (Fig. 5) and thereby its nutritional value. The concept that bacterial semiochemicals guide the foraging behav- Acknowledgments ior of flies is supported by data showing that flies select resources (including trap baits) before they make physi- We thank Tiia Haapalainen for assistance in keying out cal contact with them, and that bacterial strains produce flies; Will Ruth for assistance with statistical analyses; airborne odorants such as dimethyl trisulfide, phenol, am- Fan Sozzi-Guo for preparing various microbiological me- monia, and indole (Fig. 5), which are well-known fly dia and providing essential equipment; Stephen Takacs´ for attractants (Ma et al., 2012; Liu et al., 2016). While graphical illustrations; Sharon Oliver for word processing; these odorants have signal character to flies, the quorum- and Joe Falk and farm personnel at the field site for their

C 2018 Institute of Zoology, Chinese Academy of Sciences, 00, 1–10 The fly factor is microbe mediated 9 patience and interest in the project. This research was sup- Hendrichs, J., Cooley, S.S. and Prokopy, R.J. (1992) Post-feeding ported by a Natural Sciences and Engineering Research bubbling behaviour in fluid-feeding Diptera: concentration of Council of Canada (NSERC)—Industrial Research Chair crop contents by oral evaporative excess water. Physiological to GG, with Scotts Canada Ltd. as the industrial sponsor. Entomology, 17, 153–161. Holl, M.V. and Gries, G. (2018) Studying the “fly factor” phenomenon and its underlying mechanisms in house flies, Disclosure Musca domestica. Insect Science, 25, 137–147. Ibarra Jimenez, S., Carroll, C., Babcock, T., Derstine, N., The authors declare no conflict of interest. Hadwin, A., Moore, M. et al. (2018) Yeasts harbored by vespine wasps in the Pacific Northwest. Environmental Ento- References mology, 46, 217–225. Judd, G. and Borden, J.H. (1992) Aggregated oviposition in Acree, F., Davis, P.L., Spear, S.F., LaBrecque, G.C. and Wilson, Delia antiqua (Meigen): A case for mediation by semiochem- H.G. (1959) Nature of the attractant in sucrose fed on by icals. Journal of Chemical Ecology, 18, 621–635. house flies. Journal of Economic Entomology, 52, 981–985. Junqueira, A.C.M., Ratan, A., Acerbi, E., Drautz-Moses, D.I., Barnhart, C.S. and Chadwick, L.E. (1953) A “fly factor” in Premkrishnan, B.N.V., Costea, P.I. et al. (2017) The micro- attractant studies. Science, 117, 104–105. biomes of blowflies and houseflies as bacterial transmission Danchin, E. and Wagner, R.H. (1997) The evolution of colonial- reservoirs. Scientific Reports, 7, 16324. ity: the emergence of new perspectives. Trends in Ecology & Kearns, D.B. (2010) A field guide to bacterial swarming motility. Evolution, 12, 342–347. Nature Reviews Microbiology, 8, 634–644. Brodie, B., Gries, R., Martins, A., VanLaerhoven, S. and Lam, K., Babor, D., Duthie, B., Babor, E.M., Moore, M. and Gries, G. (2014) Bimodal cue complex signifies suitable Gries, G. (2007) Proliferating bacterial symbionts on house oviposition sites to gravid females of the common green fly eggs affect oviposition behaviour of adult flies. Animal bottle fly. Entomologia Experimentalis et Applicata, 153, Behaviour, 74, 81–92. 114–127. Largeron, M., Chiaroni, A. and Fleury, M.-B. (2008) Envi- Brodie, B.S., Babcock, T., Gries, R., Benn, A. and Gries, G. ronmentally friendly chemoselective oxidation of primary (2016) Acquired smell? Mature females of the common green aliphatic amines by using a biomimetic electrocatalytic sys- bottle fly shift semiochemical preferences from feces feeding tem. Chemistry – A European Journal, 14, A96–1003. sites to carrion oviposition sites. Journal of Chemical Ecol- Lee, J.H. and Lee, J. (2010) Indole as an intercellular signal ogy, 42, 40–50. in microbial communities. FEMS Microbiology Reviews, 34, Brodie, B.S., Wong, W.H.L., Vanlaerhoven, S. and Gries, G. 426–444. (2015) Is aggregated oviposition by the blow flies Lucilia Lenth, R. (2018). emmeans: Estimated Marginal Means, aka sericata and Phormia regina (Diptera: Calliphoridae) really Least-Squares Means. R package version 1.1. (https://CRAN. pheromone-mediated? Insect Science, 22, 651–660. R-project.org/package=emmeans; accessed 8 August 2018). Chaudhury, M.F., Skoda, S.R., Sagel, A. and Welch, J.B. Lihoreau, M. and Rivault, C. (2011) Local enhancement pro- (2010) Volatiles emitted from eight wound-isolated bacte- motes cockroach feeding aggregations. PLoS ONE, 6(7), ria differentially attract gravid screwworms (Diptera: Cal- e22048. liphoridae) to oviposit. Journal of Medical Entomology, 47, Liu, W., Longnecker, M., Tarone, A.M. and Tomberlin, J.K. 349–354. (2016) Responses of Lucilia sericata (Diptera: Calliphori- Cosse,´ A.A. and Baker, T.C. (1996) House flies and pig manure dae) to compounds from microbial decomposition of larval volatiles: wind tunnel behavioral studies and electrophysio- resources. Animal Behaviour, 115, 217–225. logical evaluations. Journal of Agricultural Entomology, 13, Ma, Q., Fonseca, A., Liu, W., Fields, A.T., Pimsler, M.L., 301–317. Spindola, A.F. et al. (2012) Proteus mirabilis interking- Eddy, G.W., DeVaney, J.A. and Handke, B.D. (1975) Re- dom swarming signals attract blow flies. ISME Journal,6, sponse of the adult screwworm (Diptera: Calliphoridae) to 1356–1366. bacteria-induced and incubated bovine blood in olfactome- MacFaddin, J.F. (2000) Biochemical tests for identification of ter and oviposition tests. Journal of Medical Entomology, 12, medical bacteria 3rd ed. (ed. McGrew, Lawrence), Lippincott 397–381. Williams & Wilkins, Philadelphia, PA, USA Gupta, A.K., Nayduch, D., Verma, P., Shah, B., Ghate, H.V., Moriya, K., Fujibayashi, T., Yoshihara, T., Matsuda, A., Sumi, Patole, M.S. et al. (2012) Phylogenetic characterization of N., Umezaki, N. et al. (1999) Verotoxin-producing Es- bacteria in the gut of house flies (Musca domestica L.). FEMS cherichia coli O157:H7 carried by the housefly in Japan. Microbiology Ecology, 79, 581–593. Medical and Veterinary Entomology, 13, 214–216.

C 2018 Institute of Zoology, Chinese Academy of Sciences, 00, 1–10 10 Y. Uriel et al.

Mulla, M.S., Hwang, Y.S. and Axelrod, H. (1977) Attractants Stoffolano, J.G., Schauber, E., Yin, C.M., Tillman, J.A. and for synanthropic flies: chemical attractants for domestic flies. Blomquist, G.J. (1997) Cuticular hydrocarbons and copula- Journal of Economic Entomology, 70, 644–648. tory behaviour in Phormia regina (Meigen). Journal of Insect Pidiyar, V.J., Jangid K., Patole, M.S. and Shouche Y.S. (2004) Physiology, 43, 1065–1076. Studies on cultured and uncultures microbiota of wild Culex Sturgill, G. and Rather, P.N. (2004) Evidence that putrescine acts quinquefasciatus mosquito midgut based on 16S riboso- as an extracellular signal required for swarming in Proteus mal RNA gene analysis. The American Journal of Tropical mirabilis. Molecular Microbiology, 51, 437–446. Medicine and Hygiene, 70, 597–603. Tomberlin, J.K., Crippen, T.L., Tarone, A.M., Singh, B., Adams, Ponnusamy, L., Xu, N., Nojima S., Wesson, D.M., Schal, C. and K., Rezenom, Y.H. et al. (2012) Interkingdom responses of Apperson, C.S. (2008) Identification of bacteria and bacteria- flies to bacteria mediated by fly physiology and bacterial quo- associated chemical cues that mediate oviposition site prefer- rum sensing. Animal Behavior, 84, 1449–1456. ences by Aedes aegypti. Proceedings of the National Academy Tomberlin, J.K., Crippen, T.L., Tarone, A.M., Chaudhury, M., of Sciences USA, 105, 9262–9267. Singh, B., Cammack, J. et al. (2017) A review of bacterial in- Richards, C.H. (1922) Is the house-fly in its natural environment teractions with blow flies (Diptera: Calliphoridae) of medical, attracted to carbon dioxide? Journal of Economic Entomol- veterinary, and forensic importance. Annals of the Entomol- ogy, 15, 425–430. ogy Society of America, 110, 19–36. Sambrook, J., Fritsch, E.F. and Maniatis, T. (1989) Molecular Verstraeten, N., Braeken, K., Debkumari, B., Fauvert, M., Cloning: A Laboratory Manual 2nd ed. (eds. Ausubel, F.M., Fransaer, J., Vermant, J. et al. (2008) Living on a surface: Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, swarming and biofilm formation. Trends in Microbiology, J.A. and Struth, K.), Cold Springs Harbor Laboratory Press, 16, 496–506. New York, NY, USA. Wieting, J.O.G. and Hoskins, W.M. (1939) The olfactory re- Schneider, J. and Wendisch, V.F. (2011) Biotechnological pro- sponse of flies in a new type of insect olfactometer. Journal duction of polyamines by bacteria: recent achievements and of Economic Entomology, 32, 24–29. future perspectives. Applied Microbiology and Biotechnol- Yamaguchi, J. and Takeda, T. (1992) Oxidation of amines with

ogy, 91, 17–30. CuBr2-LiOBu. Chemistry Letters, 21, 1933–1936. Singh, B., Crippen, T.L., Zheng, L., Fields, A.T., Yu, Z., Ma, Zheng, L., Crippen, T.L., Holmes, L., Singh, B., Pimsler, M., Q. et al. (2015) A metagenomic assessment of the bacteria Benbow, M.E. et al. (2013) Bacteria mediate oviposition by associated with Lucilia sericata and Lucilia cuprina (Diptera: the black soldier fly, Hermetia illucens (L.), (Diptera: Stra- Calliphoridae). Applied Microbiology and Biotechnology, 99, tiomyidae). Scientific Reports, 3, 2563. 869–883. Staley, J.T. and Konopka, A. (1985) Measurements of in situ activities of non photosynthetic microorganisms in aquatic Manuscript received May 4, 2018 and terrestrial habitats. Annual Review of Microbiology, 39, Final version received June 25, 2018 321–346. Accepted July 8, 2018

C 2018 Institute of Zoology, Chinese Academy of Sciences, 00, 1–10