Physiological and molecular correlates of the screwworm fly attraction to wound and animal odors

The screwworm fly, Cochliomyia hominivorax (Coquerel), was successfully eradicated from the United States by the sterile insect technique (SIT). However, recent detection of these flies in the Florida Keys, and increased risk of introductions to the other areas warrant novel tools for management of the flies. Surveillance, a key component of screwworm control programs, utilizes traps baited with rotting liver or a blend of synthetic chemicals such as swormlure-4. In this work, we evaluated the olfactory physiology of the screwworm fly and compared it with the non-obligate ectoparasitic secondary screwworm flies, C. macellaria, that invade necrotic wound and feed on dead tissue. These two species occur in geographically overlapping regions. C. macellaria, along with other blowflies such as the exotic C. megacephala, greatly outnumber C. hominivorax in the existing monitoring traps. Olfactory responses to swormlure-4 constituents between sex and mating status (mated vs unmated) in both species were recorded and compared. Overall, responses measured by the antennograms offered insights into the comparative olfactory physiology of the two fly species. We also present detailed analyses of the antennal transcriptome by RNA-Sequencing that reveal significant differences between male and female screwworm flies. The differential expression patterns were confirmed by quantitative PCR. Taken together, this integrated study provides insights into the physiological and molecular correlates of the screwworm’s attraction to wounds, and identifies molecular targets that will aid in the development of odorant-based fly management strategies.


Results
Olfactory responses to swormlure constituents. Antennal response of C. hominivorax and C. macellaria to the components of swormlure-4 8 were measured using the electroantennogram (EAG) technique. In addition, we tested the response of dimethyl trisulfide, a recently described attractive odorant for the screwworm 33 . These compounds elicited robust and highly reproducible EAG responses in both species (Fig. 1). Two compounds, acetic acid and butyric acid, elicited both hyperpolarization and depolarization in both species and therefore were omitted from subsequent statistical analyses. Majority of the antennal responses to tested chemicals-six of the nine-were not significantly different between sexes, or females' mating status. However, antennal response by C. hominivorax to two compounds, benzoic acid and phenol, differed between sex and/or mating status (Fig. 2a). Response of unmated females to benzoic acid was smaller than that of males (p = 0.023), while the response of unmated females to phenol was larger than that of mated females (p = 0.012) (Fig. 2a).
The antennal response of C. macellaria to three compounds-2-butanol, dimethyl disulfide and valeric aciddiffered among sex and/or mating status, out of the 9 chemicals tested (Fig. 2b). The response of C. macellaria unmated females and males to 2-butanol was larger than that of mated females (p = 0.023), while the response of unmated females to dimethyl disulfide was smaller than that of mated females (p = 0.023), and the response to valeric acid was larger in unmated females compared to males (p = 0.028) (Fig. 2b). Principal component analysis (PCA) of antennal responses to the nine compounds listed in Fig. 2a and b separated C. hominivorax and C. macellaria by PC1 (p ≤ 0.001, Fig. 2d); however, intraspecies differences based on sex and/or mating status were not obvious in the PCA plot (Fig. 2c). PC1 captured the most variation (39.5%). The top three PC1 loading Scientific Reports | (2020) 10:20771 | https://doi.org/10.1038/s41598-020-77541-w www.nature.com/scientificreports/ factors (weighing negatively) were 2-methyl-1-propanol, 2-butanol and dimethyl trisulfide, and the top three positively weighing compounds were indole, p-cresol and phenol.
Male and female antennal transcriptomes of C. hominivorax. A total of 10,442 genes were expressed in the antennae (Supplemental file 1), of which 114 were annotated as chemosensory genes (55 ORs,15 GRs,21 IRs and 23 OBPs). It is worth mentioning that these numbers are lower than those we reported in whole genome annotation 30 , and simply reflect the fact that some of these genes are expressed in other tissues such as legs and mouth parts, and/or at different developmental stages. A total of 85 genes were differentially expressed (DE) between the two sexes ( Fig. 3a): 34 were higher in males (male-biased) and 51 were higher in females (femalebiased). Twenty-seven of the 85 differentially expressed genes (31.8%) were chemosensory genes (16 ORs, 3 IRs and 8 OBPs) (Fig. 3a). We next explored sex-specific differences in chemosensation by conducting principle component analysis (PCA) of all 114 chemosensory genes. PCA based on gene expression of the chemosensory genes showed a clear separation of the sexes by PC1, which explained 49.5% of the variation (Fig. 3b). Validation by qRT-PCR using independently isolated RNA of six DE ORs-three each male-biased (ChomOr63, ChomOr13 and ChomOr74) and female-biased (ChomOr52, ChomOr62 and ChomOr7)-expression corroborated results based on RNA-seq (Fig. 3c).

Figure 1.
Antennal responses of the primary (left) and secondary screwworm flies to the constituent odorants of swormlure-4 as measured by electroantennography. Traces are typical EAG signals depicting downward deflections, except acetic acid and butanoic acid traces from primary screwworm flies. Scale bars are 500 ms (horizontal; also represents the onset of the stimulus delivery) and 1 mV (vertical).
Three of the 21 IRs expressed in the antennae were DE between sexes, all being female-biased (Fig. 5a). The estimated IR phylogeny suggests that these are closely related to their namesake in D. melanogaster (Fig. 5b). In D. melanogaster, Ir75a detects several carboxylic acids (acetic, butyric and propionic) 40 , while Ir76a detects amines such as pyrrolidine and phenylethylamine 41 (Fig. 5c). The types of IRs expressed in C. hominivorax were similar to those expressed in D. melanogaster antennae, with the following exceptions. ChomIr31a, ChomIr60a and ChomIr68a were not expressed in C. hominivorax but are commonly expressed in D. melanogaster, while www.nature.com/scientificreports/ ChomIr10a was expressed in C. hominivorax but is not typically expressed in D. melanogaster antennae 42 . Of the OBPs expressed in the antennae, six were DE between the sexes, with three being female-biased and three being male-biased (Fig. 6a). Five of these are closely related to D. melanogaster OBPs based on the estimated phylogeny (Fig. 6b). Thus far, a role in olfaction has been described only for DmelObp49a, where it was shown to be associated with bitter taste inhibition to sugar feeding 43 (Fig. 6c). None of the 15 GRs expressed in C. hominivorax antennae were DE between sexes. The antennal GRs included three putative CO 2 receptors, of which two (ChomGr2 and ChomGr3) were among the most highly expressed GRs, while the third (ChomGr1) was expressed at a much lower level (Fig. S1a,b). All six putative sugar receptors (ChomGr4-Gr9) were expressed in the antennae (Fig. S1a,b). The estimated GR phylogeny suggests ChomGr25 is closely related to DmelGr57a (Fig. S1b), which is associated with bitter taste in D. melanogaster 45 . These results are similar to findings in D. melanogaster in which two CO 2 receptors and five sugar receptors were expressed in the antennae 46 . Function of the remaining five GRs is difficult to predict due to their divergence from D. melanogaster GRs.

Discussion
We utilized a multifaceted approach to address knowledge gaps in screwworm fly olfactory physiology, with the aim of developing more effective and specific attractants for screwworm fly control strategies. This was the first study of its kind in which olfactory responses of mated and unmated female C. hominivorax or C. macellaria were compared. Furthermore, we report for the first time quantitative analysis of the antennal transcriptome in male and female C. hominivorax.
Antennal responses of mated and unmated female C. hominivorax to swormlure-4 constituents were similar, except for phenol, which elicited greater EAG signal in unmated females. None of the tested chemicals elicited a   34,35 , while one (Or63) is closely related to Or85f, which has been implicated in avoidance of the parasitoid volatiles actinidine and nepetalactol 37 . Two of the female-biased ORs are closely related to the 9-tricosene (Or7a), and methyl laurate, methyl myristate and methyl palmitate (Or88a) receptors in D. melanogaster 38,39 , all of which are implicated as fly pheromones. www.nature.com/scientificreports/ greater response in mated females. This is somewhat unexpected since the swormlure constituents were mostly derived from the analysis of bovine blood and odors associated with decomposition of meat protein and fatty acids representing an oviposition substrate for screwworm flies 15 . In C. macellaria, however, dimethyl disulfide elicited a larger physiological response in mated flies compared to unmated females. An additional use for a lure attracting primarily mated females, other than surveillance, is the potential for the development of a lethal oviposition trap for population control. The utility of such a trap has been demonstrated for mosquito species 47 , and remains to be for explored for control of screwworm. There were no significant differences in the antennal responses to the tested constituents when males were compared with mated and unmated female C. hominivorax or C. macellaria. This is in contrast to a report where p-cresol (4-methylphenol) and dimethyl disulfide elicited a greater physiological response in males 48 . Field studies have differed in their proportion of males and females collected in traps baited with swormlure-4, with studies ranging from no sex bias 49 to strong female-bias 8,29,50 . It was demonstrated during the development of the first synthetic lure (swormlure) that attraction was greater to the blend than to the individual constituents 15 , therefore, it is possible that the blend (swormlure-4) elicits a sex biased response while the individual constituents do not.
Multivariate analysis (PCA) suggests C. hominivorax and C. macellaria differed in their antennal responses to the swormlure constituents, even though sex or physiological state (virgin vs. gravid) differences were not that apparent within a species. However, this should be interpreted with caution: EAG studies for C. macellaria were performed in the laboratory at the University of Kentucky wherein optimum recording condition, such as stable temperature, humidity and noise-free electrical contacts are observed. Antennal recordings of C. hominivorax conducted at COPEG didn't have all these optimal conditions. Consequently, we avoided direct comparison of EAG responses of C. hominivorax with C. macellaria. Due to their close evolutionary relationship yet different oviposition substrates preference (live animal wounds vs carrion or necrotic tissue), comparative chemical ecology and molecular olfaction studies of C. hominivorax and C. macellaria could improve the species-specificity of synthetic attractants for the screwworm fly and yield insights into the evolution of parasitism in Cochliomyia in the future.
In contrast to our EAG analysis of C. hominivorax in which there were no major sex-specific differences in antennal responses elicited by the tested chemicals, antennal transcriptome analysis revealed 29% of the expressed ORs as sex-biased. The simplest explanation for the disparity between EAG responses and gene expression analysis is that the EAGs are only proximate summation of the individual olfactory receptor neuron (ORN) responses 51 , and do not faithfully reflect the response of all chemosensory genes. Recent evidence also suggests variation in the transcriptional profiles of the chemosensory genes in natural mosquito populations 52 . Another likely explanation for absence of any sex-based correlation between the transcriptional profiles (Fig. 3) and EAG responses (Fig. 2) is that the ORs that are major contributors for the sexually dimorphic transcription profiles do not significantly contribute to the EAG amplitudes measured for the swormlure-4 constituents tested here. Our phylogenetic analysis of the olfactory genes groups nine of the 16 sex-biased C. hominivorax ORs with D. melanogaster receptors associated with pheromone sensing 35,38,39 . In Drosophila, trichoid sensilla respond to pheromone related odors, whereas most of the food/host odors are detected by basiconic sensilla 54,55 . Cho-mOr63 (male-biased) is closely related to DmelOr85f, which was shown to be associated with innate avoidance of actinidine and nepetalactol, volatiles associated with the Drosophila parasitoids in the genus Leptopilina 37 (Fig. 4b,c). ChomOr19 and ChomOr62 (female-biased) are closely related to DmelOr7a and DmelOr88a, respectively (Fig. 4b). DmelOr7a was shown to be necessary for aggregation and oviposition site selection via 9-tricosine signaling 39 , while DmelOr88a was required for short-range attraction of both sexes to the fly-produced odors methyl laurate, methyl myristate and methyl palmitate 38 (Fig. 4c). The remaining seven sex biased ORs had either no close phylogenetic relationship or their function has not yet been well characterized in D. melanogaster. Previous studies have established a role for chemical sexual communication in C. hominivorax, implicating male-produced 56 and female-produced 57,58 substrates such as cuticular hydrocarbons (CHCs). However, despite the identification of several CHCs 59-61 , their behavioral significance has yet to be realized. With the availability screwworm genome 30 , and specifically the development of genetic knockout techniques in these flies such as CRISPR/Cas9 62 , we aim to understand these behaviors driven by chemosensory genes. Identification of the ligands for the male-and female-biased receptors we describe here could identify chemicals for the development of sex-specific attractants.
The contribution of antennal IR expression on the EAG responses is unclear. These genes are primarily expressed in coeloconic sensilla 41,42 . Two of the three female-biased and highly-expressed IRs in the C. hominivorax antennal transcriptome (ChomIr75a-2 and ChomIr75a-3) are closely related to DmelIr75a, which responds to the short chain carboxylic acids 40 . Electroantennograms represent an approximate global olfactory responses from the antenna depicting the overall response dynamics and amplitude resulting from multiple ORNs expressing different families of chemosensory receptor genes, such as ORs, IRs or GRs 63,64 . As we noted earlier, acetic acid and butyric acid induced occasional hyperpolarizations. Short chain carboxylic acids, especially at high doses, often elicit upward deflection in EAG recordings across a range of insects 65 , and this has been ascribed to the acidity of these compounds and/or sometimes, as an artifact electrode potentials resulting from interaction of acids with the electrodes 66 . Our conclusion of the acid induced hyperpolarization, based on multiple measurements across both sexs and species, is that the measured signals are not artifacts.
Monitoring for the presence of C. hominivorax is an important component of screwworm control and eradication programs 67 . Though various measures are utilized for monitoring, synthetic attractants (e.g. swormlure) are particularly appealing due to the ease and expense at which they can be employed compared to alternative attractants, such as rotten beef liver. However, gaps in our knowledge have prevented advancement of synthetic attractant strategies for the screwworm fly. Improvement in species specificity and the development of a malespecific and gravid-female specific attractants is highly desirable. Here, we used a multifaceted approach in which we integrated neurophysiology, molecular and phylogenetic methods to gain a better understanding of C. hominivorax biology, toward the development of novel attractants for control of screwworm. We are in the process of functional characterization of the promising chemosensory genes we identified here based on the differential expression pattern and their phylogenetic relationship to the D. melanogaster orthologues that have been fully characterized. Prospecting the cognate ligands from the natural milieus such as open wounds and conspecifics will lead to the isolation and identification of chemostimuli that will be tested for the behavioral response, and eventually as odor baits in the surveillance traps.

Methods
Flies. The J06 strain of C. hominivorax was used for antennal transcriptome analysis and electroantennography (EAG). The J06 wild type strain of C. hominivorax was originally collected in Jamaica in 2006, and this strain was being used at the COPEG in Panama for SIT release at the time of the study.Flies were reared at the COPEG biosecurity plant in Panama using methods described previously 68  Weekly, adult fly cages were provided with 10 g of raw beef liver as a protein source to optimize ovary development. Two-week-old flies were provided a cup containing 10 g of raw liver loosely covered with a moist paper towel to encourage oviposition. Collected eggs were partitioned into pea sized (10 mm) allotments for transfer to the prepared larval medium consisting of 750 ml dry Purina Cat Chow Complete (Nestle Purina PetCare Company, St. Louis, MO) soaked in 400 ml of water. In a well-ventilated room, the larval medium bucket was placed on a rack at a 35° angle, the eggs added and the bucket covered with a screened lid. Pupae were harvested in 7 to 8 days.
EAG recordings. Mounting of C. hominivorax for EAG was conducted by excising the head of the fly (3-6 days post-eclosion) with microscissors and placing it on an antenna holder (Syntech, Germany) with the base of the head on one side and the tip of both antennae contacting the other. To improve electrical conduction and reduce desiccation, Spectra 360 Electrode Gel (Parker Laboratories) was applied at the contacts between the holder and the fly (base of head and tip of antennae). We note that due to biosecurity measures, these recordings could only be conducted inside the COPEG production facility, which is not optimal for fine electrophysiological measurements. Mounting of C. macellaria for EAG was similar to fruit flies recordings we routinely employ 69 with a few modifications. A fly (3-8 days post-eclosion) was inserted in a 1000 μl pipette tip (USA Scientific Inc.)

Scientific Reports
| (2020) 10:20771 | https://doi.org/10.1038/s41598-020-77541-w www.nature.com/scientificreports/ with approximately one-third of the head protruding from the tip. The tips of both antennae were placed in a 1 mm borosilicate capillary (World Precision Instruments, USA) containing sensillum lymph ringer 65 , while a ground was placed into the eye. Signals were amplified and recorded using an IDAC2-USB box (Syntech) with a 0.05 Hz low cutoff. Each recording was a total of 10 s in length with a set 1 s pre-trigger. A CS-55 Stimulus Controller (Syntech) was used to supply a charcoal filtered and humidified continuous air stream (0.8 l/min) delivered via a Teflon tube. A stimulus pulse (0.8 l/min) was added to the air stream for 0.5 s. Recordings were analyzed with EAG Pro version 1.1 software (Syntech). Chemicals used for stimulus and solvent were of high purity: 1-octen-3-ol (98% purity) 2-butanol (analytical standard), 2-methyl-1-propanol (analytical standard), acetic acid (99.5%), benzoic acid (analytical standard), butyric acid (analytical standard), dimethyl disulfide (analytical standard), dimethyl trisulfide (98%), indole (98%), p-cresol (≥ 99%), phenol (≥ 99%) and valeric acid (analytical standard). Chemicals were dissolved in dichloromethane (DCM ≥ 98%) to make a stock solution of 100 mg/ml (− 1). A 1:10 dilution (− 2) of the stock solution in DCM was used as stimulus. A 20 µl of stimulus solution was loaded onto a filter paper strip. After allowing the solvent to evaporate for 10-15 s, the strip was placed in a 5 ml polypropylene syringe for delivery onto the EAG preparation. Each recording was first baseline normalized, then smoothed with a 200 ms running average, amplitude determined and the value of the appropriate blank (solvent) control subtracted.
The antennal response for each compound is represented as the proportion of the summed responses of each individual to all compounds. Statistical tests for difference in antennal response between sexes and/or physiological state (virgin and gravid), and principal component analysis (PCA) were conducted using SigmaPlot v14.0. Data were initially assessed and passed tests for normality (Shapiro-Wilk) and equal variance (Brown-Forsythe) prior to implementing a one-way ANOVA and Tukey post-hoc test (when ANOVA p-value < 0.05). Principal component analysis was conducted with the covariance method using the average eigenvalue for component selection. The Mann-Whitney rank sum test was used to test for difference between principal component scores for PC1 in Fig. 2d.
Antennal transcriptome analysis. The methods used for extraction and quantification of total RNA, and preparation of cDNA libraries are described in Scott et al. 30 . In brief, antennal RNA was extracted from frozen samples, and RNA integrity, purity, and concentration were assessed using an Agilent 2100 Bioanalyzer. Purification of mRNA was performed using the oligo-dT beads provided in the NEBNExt Poly (A) mRNA Magnetic Isolation Module. Complementary DNA libraries for Illumina sequencing were constructed using the NEBNext Ultra Directional RNA Library Prep Kit (NEB) and NEBNext Mulitplex Oligos for Illumina (NEB) using the manufacturer-specified protocol. The amplified library fragments were purified and checked for quality and final concentration using an Agilent 2200 Tapestation with a High Sensitivity DNA chip (Agilent Technologies, USA) and a Qubit fluorometer (ThermoFisher, USA). The quantified cDNA libraries were pooled in equimolar amounts for clustering and sequencing on an Illumina HiSeq 2500 DNA sequencer, utilizing a 125 bp single-end sequencing reagent kit (Illumina, USA). Sequencing depth ranged from 8.74 to 13.06 million mapped reads per library (mean = 10.45). Sequence reads were mapped to the C. hominivorax genome assembly using Hisat2 v2.1.0 70 . Read counts were obtained using HTSeq-count (union mode) in HTSeq v0.11.2 71 . The quasi-likelihood F-test and false discovery rate (fdr) correction for multiple tests in EdgeR was used for differential expression 72,73 . Data were converted from CPM (counts per million reads) to RPKM (reads per kilo base per million mapped reads) following statistical analysis. SigmaPlot v14.0 was used for PCA of chemosensory gene expression (RPKM) using the correlation method and the average eigenvalue for component selection.
RT-qPCR validation of candidate C. hominivorax genes. RNA was isolated following the Direct-zol protocol (Zymo Research, Irvine, CA) from five pairs of male and female adult fly antenna that were shipped from COPEG, Panama in RNA-shield (Zymo Research, Irvine, CA). The quantity and quality of the isolated RNA was assessed using a NanoDrop. First-strand cDNA was synthesized from 300 ng of total RNA per sample using the Superscript II protocol (Thermo Fisher, Carlsbad, CA). cDNA for each sample was diluted to 20 ng/µl for subsequent qPCR reactions. Actin and RPS17 served as endogenous control genes for qPCR since these were shown to be the top and most stably expressed genes within the members of Calliphoridae 74 . Each 10 µl PCR reaction included 2 µl of cDNA template (20 ng/µl), 5 µl of SYBR GRN master mix, 1 µl of primer (0.5 µl F and 0.5 µl R) and 2 µl of sterile water. PCR cycling conditions were 95 °C (3 min) and 40 cycles of 95 °C (10 s) and 60 °C for 30 s. For each gene, a standard curve was derived to assess the amplification efficiency of the primers, melting curve to assess the specificity of the primers and included three biological replicates. Differential expression analysis was based on the well establish method 75 .
Phylogenetic analysis of chemosensory genes. Phylogenetic relationships among C. hominivorax 30 and D. melanogaster 40,76,77 chemosensory genes were estimated by first conducting a multiple protein alignment for each gene family using default parameters in ClustalX v2.1 78 . Alignments were visually inspected and then used for maximum likelihood estimation in RAxML v8.2.11 with the PROTGAMMAAUTO model option and 500 bootstrap replications 79 . Figures were generated using the interactive Tree of Life (iTOL) v4 software 36 (https ://itol.embl.de/).