Background

Sexual selection is a powerful evolutionary force that can drive species divergence [1–4]. Although many studies focus on how organisms choose mates during pre-mating sexual selection, the process is not limited to courtship, but rather occurs across multiple time points before, during, and after copulation [5–7]. Post-mating sexual selection has proven challenging to study because it can involve interactions between the female reproductive tract and the male ejaculate on a molecular level. Such interactions include male-male sperm competition, sexual conflict involving male and female proteins, and cryptic female choice. Sexual conflict is of particular interest because it arises from the divergent reproductive interests of males and females and thus may represent an important component of post-mating interactions [8]. Many mechanisms of post-mating sexual selection involve co-evolutionary arms races between the sexes, a process that can lead to rapid trait evolution within [8, 9] or divergence between populations [10–12]. Differences in reproductive traits between populations can quickly result in post-mating, pre-zygotic barriers [13], potentially playing a powerful role in species formation.

Most work on post-mating, pre-zygotic barriers has focused on interactions between the female reproductive tract and male sperm. Specifically, cryptic female choice has been identified as a possible mechanism for conspecific sperm precedence, in which multiply mated females produce more offspring sired by conspecific rather than heterospecific mates [14–16]. Conspecific sperm precedence is widespread and has been demonstrated in many insect species, including fruit flies, ground crickets and flour beetles [14, 16–19]. Of several possible mechanisms underlying conspecific sperm precedence in Drosophila mauritiana and Drosophila simulans, biased sperm use by females was found to be a key determinant [20]. Female Drosophila are able to favor conspecific males by preferentially storing sperm in separate storage organs [20]. Although multiply mated females seem to be able to bias paternity of their offspring, the nature of the interactions between male and female reproductive proteins that might lead to such differential sperm use is unclear [21, 22].

After mating, the female reproductive tract interacts not only with male sperm, but also with seminal fluid. In many taxa, male reproductive proteins are produced in accessory glands or the ejaculatory duct, which are then transferred to the female as components of the male ejaculate. Collectively, these non-sperm components of the ejaculate are called seminal fluid proteins and they are quite numerous [23]. In fruit flies, Drosophila melanogaster males produce over 100 seminal proteins that are transferred to females during mating [24–26]. These proteins have profound effects on female behavior and physiology, including changes in lifespan, ovulation, feeding habits and sperm storage patterns [14, 24, 26]. Not only do male reproductive proteins have important effects on females, they are potentially powerful drivers of post-mating, pre-zygotic reproductive isolation because many of them evolve rapidly [22]. Seminal fluid proteins have been comprehensively characterized in several insect taxa, including fruit flies, mosquitoes, honeybees, crickets, flour beetles, butterflies, and bedbugs [24, 27–35]. In contrast, we know very little about the many possible interacting female reproductive proteins for any one species.

Although the reproductive tracts of female insects also contain secretory tissue [36], to date female reproductive genes have been comprehensively studied in very few taxa including mosquitos, fruit flies, and honeybees [37–44]. Unsurprisingly, female reproductive genes have been best characterized in Drosophila species including: D. melanogaster, D. simulans, D. arizonae and D. mojavensis[38–43]. Many of these investigations have identified proteases and protease inhibitors, as well as genes related to muscle activity, immune response, and energy metabolism in female reproductive tracts [38–43]. Genes with these functions are predicted to mediate interactions with male ejaculate after mating. Indeed, muscle activity is a key component of female-mediated sperm storage and ejaculate processing [45, 46], while proteases and protease inhibitors have been shown to be required for activation of ovulation-inducing seminal fluid proteins in D. melanogaster[47]. Furthermore, immune and energy metabolism genes appear to be important for the demands of egg production and oviposition or to protect females from male-introduced pathogens [38, 40, 48]. Many relevant female reproductive proteins are secreted from female tissue or are transmembrane, as these are likely to directly interact with male ejaculate or act as receptors for male seminal fluid proteins [42].

To reach a comprehensive understanding of the mechanisms by which male and female reproductive genes drive post-mating sexual selection and ultimately species divergence, studies of reproductive proteins must consider organisms with different mating systems and different levels of sexual conflict. As the second largest insect order with ~170,000 known species, moths and butterflies comprising the order Lepidoptera are ideal for study because the degree of multiple mating by females, and thus the opportunity for sexual conflict, positively correlates with speciation rate [49, 50]. However, very little is known about male and female reproductive genes in Lepidoptera. Two comprehensive studies, both in Heliconid butterflies, have identified male reproductive proteins in lepidopterans [34, 35]; however, researchers have yet to identify female genes from any structure in the lepidopteran female reproductive tract.

Most male lepidopterans transfer their ejaculate in a package called a spermatophore (Figure 1a) [36, 49]. Although produced by males, the spermatophore is actually formed inside a large, sac-like structure inside the female called the bursa copulatrix (Figure 1b). Inside the bursa copulatrix, the spermatophore is broken open by the signum, a chitinized structure embedded in the muscular wall of the bursa copulatrix, and both sperm and male seminal fluid proteins are released into the female reproductive tract; however, the spermatophore remains in the bursa copulatrix as a visible structure for the entirety of the females life [36, 49]. Thus, the bursa copulatrix represents a strong candidate arena for the resolution of sexual conflict and the origin of post-mating, pre-zygotic isolation. In general, males that prevent females from remating will achieve greater paternity success. In species where this is true, male traits will evolve that delay female remating. For example, males that transfer larger spermatophores are able to delay female remating for longer in some lepidopterans [51, 52] and in D. melanogaster, identified male reproductive proteins act to reduce female receptivity to future mates [47, 53]. On the other hand, females in many taxa gain material and genetic benefits from multiply mating [54, 55], and therefore selection will favor morphological and biochemical traits that allow females to rapidly process male spermatophores. Although recent microstructural studies in Lepidoptera suggest the bursa copulatrix could have a secretory function, studies have yet to characterize any secretions from the structure [56]. Such secretions could be important for breaking down spermatophores or for interacting with male reproductive proteins.

Figure 1
figure 1

Male and female ECB reproductive structures. a. An ECB male spermatophore transferred to a female upon mating. b. The virgin female bursa copulatrix with signum (indicated by arrow) and the bursal gland (indicated by*). Notice that even in virgins the bursal gland is filled with fluid. Scale bars represent 1 mm.

Full size image

A second structure found in some lepidopteran females that could mediate within- and between-species mating success is the bursal gland. Although patterns of evolutionary conservation remain unclear, the bursal gland is a prominent anatomical feature of the female reproductive system in the European corn borer moth, Ostrinia nubilalis[57]. The bursal gland is dorsally connected directly to the bursa copulatrix by a short duct (Figure 1b) and is approximately 0.5-0.8 mm in diameter. In virgin females the bursal gland is filled with a white, translucent fluid, which then flows into the bursa copulatrix under pressure [57]. After mating, the bursal gland is similarly filled, but with an opaque fluid [57]. The function of this gland is currently unknown, however its proximity and direct connection to the bursa copulatrix and male spermatophore suggests that the bursal gland could function during spermatophore breakdown or to secrete female reproductive proteins regulating the activity of male reproductive proteins.

Here, we use next-generation RNA sequencing to characterize gene expression in the female bursa copulatrix and bursal gland as the first step towards determining how these tissues are involved in post-mating, pre-zygotic isolation in the European corn borer moth (hereafter, “ECB”). The Z and E strains of ECB are emerging textbook models for the study of speciation [13], in which the two incipient species split approximately 75,000 to 150,000 years ago through the evolution of manifold reproductive barriers [58, 59]. Females of both strains mate multiply [57] and suffer reduced longevity after mating [60], conditions that are generally favorable for sexual conflict and the evolution of post-mating, pre-zygotic isolation. Consistent with this notion, one of seven barriers between strains, accounting for a ~30% reduction in gene flow, stems from reduced lifetime fecundity following between-strain mating. This post-mating, pre-zygotic incompatibility is asymmetric: Z-strain females that have mated with E-strain males lay significantly fewer eggs over their lifetime [59]. However, the mechanism underlying this gametic isolation is unknown [59]. By examining the transcriptome of bursa copulatrix and bursal gland reproductive tissues within and between ECB strains, we characterize candidate genes that could be contributing to the egg-laying dysfunction. Specifically, we characterize the function of the bursa copulatrix and bursal gland and identify female reproductive genes that may be involved in isolation using the following criteria: (1) putative proteins that are secreted or membrane bound, (2) an up-regulation of transcripts that aid in muscle contraction and that may assist in spermatophore breakdown or sperm transfer to storage, (3) an up-regulation of proteases and protease inhibitors that could mediate male seminal fluid protein potential, and (4) differential expression between Z and E strains.

Methods

Sample preparation and sequencing

We collected bursa copulatrix and bursal gland tissues from 2-day old adult Z- and E-strain ECB females (n = 12 per strain, Figure 1). At this stage females are reproductively mature [60, 61]. As the goal of this study was to identify the reproductive function of these two structures and characterize differences between ECB strains and not to identify genes directly affected by mating, all female tissues were collected from virgins. The following dissections were done in RNAlater (Qiagen, California). First, females were sacrificed and the bursa copulatrix and attached bursal gland were removed from an incision in the female abdomen. Next, fat body was removed from both structures and, after separating the bursal gland from the bursa copulatrix, both tissues were stored in RNAlater at -80°C. After tissue collection, total RNA was extracted from bursa copulatrix and bursal gland tissues using an RNeasy Midi kit (Qiagen, California). Bursal glands and bursa copulatrix tissues from four females were pooled by strain into three separate samples for each tissue type and each strain prior to an initial tissue homogenization step. This resulted in twelve samples, three bursal gland samples and three bursa copulatrix samples for each strain. RNA quantities were assessed using a Nanodrop and 1 μg of total RNA from each sample was used to create cDNA libraries (Illumina Truseq RNA sample preparation kit v2, San Diego, CA). To prepare samples for sequencing, mRNA was selected from each sample using poly-T-tail magnetic beads. Next, cDNA was synthesized using Superscript II (Invitrogen, Grand Island, NY) and Illumina adapters were attached to libraries for multiplexing prior to sequencing. cDNA strands were then amplified using 15 PCR cycles. Next, quality and quantity of cDNA was confirmed using a NanoDrop (Thermo Fisher Scientific Inc., Delaware) and an Agilent 2100 Bioanalyzer (Santa Clara, CA). Due to low sample quality for one Z-strain bursa copulatrix library, we did not sequence this sample, which left us with two Z-strain bursa copulatrix tissue libraries. Four Z- and E- bursal gland and bursa copulatrix cDNA libraries were multiplexed and sequenced in each of three lanes of two Illumina flow cells. Libraries were sequenced using an Illumina HiSeq2000 with 50 bp single-end reads. To evaluate tissue-biased expression, we took advantage of previously sequenced Z- and E- female thorax libraries developed during a separate project. Z-strain and E-strain female thoraxes were collected from 2 females each. Briefly, female thorax cDNA libraries were created using the SMART cDNA library 6.7 protocol (Takara Bio, Otsu, Shiga, Japan). These libraries were sequenced using an Illumina GA IIx and 40×40 bp paired-end reads with a 200 bp insert length.

Data preprocessing

After sequencing, all reads were subjected to quality control and trimming using Trimmomatic v0.17 to remove Illumina sequencing adapters and low quality reads [62]. Leading and trailing bases with a quality score < 5 were trimmed from reads and then each read was trimmed by a sliding window with a width of 4 bp and minimum average quality of 15. After adapter and quality trimming, only reads ≥ 36 bp were retained.

Although we used magnetic beads to select for mRNA, our samples still contained small amounts of mtDNA and rRNA sequences. To remove these contaminants we used the short read aligner Bowtie 2 [63]. Bowtie2 uses the Burrow-Wheeler transformation to index a reference, then searches the index until it finds an alignment for a specific read [63]. We aligned our RNAseq reads to the complete ECB mitochondrial genome (NC_003367.1), and all published ECB ribosomal sequences [AF336303.1, AF077013.1, DQ988989.1, AB568463.1, AY513653.1, JX683305.1, JX683313.1, AB568278.1, AB568276.1, AB568274.1, EU532443.1, EU532441.1, EU532439.1, EU532444.1, EU532442.1, EU532440.1, EU532438.1, AF349036.1] in the NCBI database using default parameters and removed these sequences. Identical reads were then collapsed using FastX toolkit to reduced library complexity and decrease the computational needs for transcriptome assembly [64].

De novo sequence assembly

We used the Trinity program suite to assemble all 13 tissue libraries including female bursa copulatrix and bursal gland libraries, as well as the two thorax libraries into a single assembly [65]. Trinity uses the inchworm, chrysalis, and butterfly software modules to create a de novo assembly. First, inchworm assembles reads into unique sequences. Chrysalis then clusters sequences into contiguous sequences and a de Bruijn graph is created for each cluster of contiguous sequences. Lastly, butterfly uses the de Bruijn graphs to construct transcripts. For all steps, we used settings recommended by developers, including merging the assembly at a kmer of 25 [65, 66]. Finally, following de novo assembly, we selected the longest transcript at each locus to eliminate redundancy.

Annotation

To annotate our assembled transcriptome, we used the program Blast2Go [67, 68]. First, putative homologs were identified by performing a blastx search of the entire NCBI non-redundant protein database (e-value cutoff 10-3). For all sequences with significant blast hits, four different mappings were conducted. First, BLAST accession numbers are used to find gene names and symbols from NCBI gene_info and gene2accession. Then, gene_info identifiers are used to retrieve UniProt IDs using PSD, UniProt, SwissProt, TrEMBL, RefSeq, GenPept and PDB databases. In the final two steps, BLAST accessions were searched in the dbxreftable and the gene product table of the GO database. Finally, Blast2go computed an annotation score for all possible GO terms for each sequence [67, 68].

Differential expression analysis

To identify differentially expressed sequences, we first mapped our reads back to our assembled transcriptome using Bowtie 2. For read mapping, we used the 'very sensitive’ setting in Bowtie 2 because preliminary trials indicated that this setting resulted in the most uniquely aligned reads. Differences between the SMART and TRUseq library preparation protocols could potentially lead to biases related to GC content, read length, and sequencing depth in each library. To help control for these and other possible biases, we then normalized our libraries prior to differential expression analysis using the programs EDAseq [69]. EDAseq performs within-lane normalization to account for differences in gene length and GC content and between-lane normalization to account for differences in sequencing depth [69]. Within-lane normalization uses global scaling normalization, which separates genes into equally sized bins based on GC-content and then matches different parameters of the count distribution across bins. For between- lane normalization, EDAseq uses a full-quantile normalization procedure that forces equal library sizes across lane.

After normalized read counts were obtained, we used the R package, edgeR to identify differentially expressed genes for all comparisons of interest using the normalized counts for each library (Figure 2) [70–73]. edgeR uses empirical Bayes methods to estimate gene-specific variation. As we were interested in four comparisons in particular, we used a generalized linear model approach in which we assessed differential expression with strain and tissue type as factors (Figure 2). Our model did not include an interaction term. Finally, a GLM likelihood ratio-test was used to identify differentially expressed genes [70]. Genes were considered differentially expressed if they had a false-discovery rate (FDR) of <0.01.

Figure 2
figure 2

Differential expression analysis tissue comparisons. Our generalized linear model used to calculate differentially expressed genes between E- and Z- strain ECB included the following comparisons: female thorax versus the female bursa copulatrix, female thorax versus bursal gland, Z-strain versus E-strain bursa copulatrix, and Z-strain versus E-strain bursal glands.

Full size image

Characterizing bursa copulatrix and bursal gland function

We used a three-pronged approach to characterize the reproductive function of the bursa copulatrix and the bursal gland. First, we adopted a common method to determine the specific functions of these tissues by ignoring housekeeping genes that have similar expression profiles across reproductive and non-reproductive tissues. For all of the remaining transcripts with significant expression differences between the bursa copulatrix and thorax, or between the bursal gland and thorax, gene annotations were pulled and enriched/depleted gene ontology categories were identified using a two-tailed fisher’s exact test in Blast2go with a term filter cutoff of FDR ≤ 0.05. Our entire non-redundant transcriptome containing transcripts from all three tissue types was used as the null distribution of GO categories.

Second, we identified signal peptides and transmembrane helices from the bursa copulatrix and bursal gland non-redundant transcriptome. For the purpose of identifying secreted and transmembrane proteins in the bursa copulatrix and bursal gland, we used a tblastx to remove all thorax sequences from our transcriptome. Next, to estimate predicted protein sequence from female bursa and bursal gland RNA-seq libraries, we used ESTscan [74], in which biases in hexanucleotide usage in coding versus non-coding regions and a Hidden Markov Model are used to predict protein-coding sequences. Subsequently, we identified sequences containing a secretion motif using SignalP 4.0 [75], which uses a neural network-based method to identify signal peptides. We then used TMHMM 2.0 to identify sequences with transmembrane helices [76], in which a Hidden Markov Model is used to predict integral membrane proteins.

Our last step to characterize the function of the bursa copulatrix and the bursal gland was to identify putative ECB homologs of female reproductive genes in other organisms. We obtained genes lists from studies on female reproductive genes for the following taxa: D. simulans, D. melanogaster, D. arizonae, Apis melifera and Anopheles gambiae[37, 40, 42–44, 77–79]. These studies either had the goal of identifying female reproductive genes or looked at expression changes in mated females compared to virgin females. Our search yielded a list of 2,952 contigs, which were then used as queries in a BLAST search against our transcriptome without thorax sequences using the tblastx algorithm and an e-value cuttoff of 10-5.

Comparison of E- and Z- reproductive genes

As a final approach to examine the bursal gland and bursa copulatrix for possible roles in post-mating sexual selection or post-mating, pre-zygotic isolation, we explored patterns of gene expression between Z and E strain females. Although differential expression alone could be viewed as evidence in support of a functional relationship and mechanism underlying dysfunctional inter-strain oviposition [59], here we emphasize enriched or depleted functional terms. After identifying differentially expressed genes between E and Z strain bursas and bursal glands respectively, we used a two-tailed fisher’s exact test (cutoff of FDR ≤ 0.05) to identify relevant GO categories and genes that might account for reduced fecundity after between-strain matings [67, 68].

Results

De novo sequence assembly

Single-end Illumina sequencing of 11 ECB female reproductive tissue samples yielded more than ~ 700 million raw reads. Paired-end Illumina sequencing of 2 ECB thorax samples yielded ~ 6 million raw reads. The assembled transcriptome of all 13 libraries contained 92,335 transcripts belonging to ~ 51,000 loci with a mean sequence length of 991 bp and a minimum of 201 bp (Table 1). This is likely an overestimate of the number of loci represented in our transcriptome due to de novo assembly limitations. Our mean assembled transcript lengths are greater than or equal to those reported in similar studies using the same sequencing technology [80, 81]. Prior to library normalization, log-fold change of read counts between samples differed, which can bias differential expression results. GC-content of each sample also differed prior to normalization. After normalization, gene level counts and GC-content between samples were all equal across libraries (Additional file 1: Figure S2 and Additional file 2: Figure S3).

Table 1 Assembly statistics
Full size table

Characterizing reproductive function of the bursa copulatrix and bursal gland tissue

Our first approach to examining the reproductive function of the bursa copulatrix and bursal gland was to characterize differences in gene expression between reproductive and non-reproductive tissues. A total of 2,982 transcripts were differentially expressed between the bursa copulatrix and the thorax, representing 8% of all bursa copulatrix genes, whereas 3,316 genes were differentially expressed between the bursal gland and the thorax, representing 9% of all bursal gland genes (Additional file 3: Figure S1). For gene ontology terms enriched in the bursa copulatrix, 20% corresponded to categories related to muscle activity and organization (n = 596), while the bursal gland had 3% of genes enriched for the same categories (n = 89) (Additional file 4: Table S1). The bursal gland was also enriched for 6 gene ontology categories directly related to sexual reproduction and gamete production (Additional file 4: Table S1). Of all up-regulated differentially expressed transcripts in the bursa copulatrix compared to the thorax, 2% contained a signal peptide indicating they are secreted from the bursa copulatrix and 7% contained at least one transmembrane helix, while the bursal gland and thorax comparison yielded 16% of sequences with a signal peptide and 46% with at least 1 transmembrane helix (Table 2).

Table 2 Up-regulated bursa copulatrix and bursal gland genes with secretion signal peptides or transmembrane motifs
Full size table

Next, we examined the results from our cross-species female reproductive gene comparison to identify conserved classes of female reproductive genes across insect taxa. We found 23 putative ECB homologs in D. simulans, D. melanogaster, D. arizonae, and A. melifera, but no homologous sequences in A.gambiae (Table 3). Across all four species with significant blast hits to ECB bursa copulatrix or bursal gland genes, many possessed gene ontology functions related to muscle contraction (Table 3).

Table 3 Between-species comparison of female reproductive genes
Full size table

Comparison of E- and Z- reproductive genes

To identify female reproductive genes that may be contributing to reproductive isolation between Z and E ECB strains, we searched for differentially expressed genes between strains. Between the Z and E strain bursa copulatrix tissues, there were 864 genes with significant differential expression and for bursal gland tissues we found 1,390 significantly differentially expressed genes between strains (Additional file 3: Figure S1).

Subsequently, we examined enriched and depleted gene ontology classes with genes that were differentially expressed between ECB strains in either the bursa copulatrix or the bursal gland. Here, we found 7 gene ontology categories, including proteolysis and serine-type peptidase and endopeptidase activity, which were significantly enriched in the bursa copulatrix compared to our transcriptome and 7 that were depleted (Figure 3). For the bursal gland, we found one gene ontology class, structural constituent of cuticle, that was enriched compared to our transcriptome and one gene ontology category, intracellular, that was depleted.

Figure 3
figure 3

Enriched and depleted gene ontology categories in the ECB bursa copulatrix. Fisher’s exact test results for enriched and depleted gene ontology categories found in the bursa copulatrix compared to the entire transcriptome. All enriched gene ontology categories were related to protease activity in the bursa copulatrix except for. Pink bars represent the gene ontology categories found in the entire transcriptome, while the gray bars represent the gene ontology categories found for the genes differentially expressed between E- and Z- strains of ECB.

Full size image

To further explore female reproductive genes that were differentially expressed between strains, we examined our gene ontology lists for proteases and protease inhibitors. We focused on these classes because they mediate male–female post-mating interactions in Drosophila and are rapidly evolving, which makes these proteins likely to be involved in sexual conflict [82, 83]. By manually searching annotations lists after a Fisher’s exact test was run in Blast2go, we found 44 proteases and 7 protease inhibitors with differential expression in both strains and in both tissues combined (Tables 4 and 5).

Table 4 Differentially expressed proteases and protease inhibitors in the bursa copulatrix
Full size table
Table 5 Differentially expressed proteases and protease inhibitors in the bursal gland
Full size table

Discussion

This study represents the first comprehensive characterization of female reproductive genes in any lepidopteran system. Using RNAseq, we identified female reproductive transcripts from the bursa copulatrix and bursal gland and from Z and E strain ECB and we characterized genes that may be involved in reproductive isolation between strains. The bursa copulatrix appears to act as a muscular sac, but it does not seem to secrete the variety of proteins found in the bursal gland, many of which are directly related to reproduction. We also found that most differentially expressed genes in the bursa copulatrix and many in the bursal gland were proteases, which could be important in post-mating sexual selection and post-mating, pre-zygotic barriers. These are of particular interest because in other species, proteases are known to be involved in female interactions with male sperm and male seminal fluid proteins [82].

Bursa copulatrix

As the site of initial interaction between male ejaculate and the female reproductive tract [57], the bursa copulatrix is likely to be an important arena for sexual conflict. Previous work has shown that the signum helps break open the male spermatophore [36], and within Papilionidae butterflies signum complexity correlates with the thickness of the outer covering on the spermatophore [84, 85]. However, a recent study on the microstructure of the bursa copulatrix in Tortricidae moths identified pores that were suggested to perform a secretory function [56], such as processing the spermatophore.

The transcriptome of the bursa copulatrix would appear to support its role as a site for the mechanical disruption of the spermatophore. Up-regulated genes in the bursa copultrix compared to the female thorax were statistically significantly enriched for GO classes related to muscle structure or activity (Additional file 4: Table S1), and many of the female reproductive genes in flies and honey bee showing homology with ECB sequences had muscle contraction functions (Table 3). For example, bursa copulatrix transcripts comp6763_c0_seq1, comp26288_c0_seq1, and comp6703_c0_seq1 had strong hits to actin 57B, myosin heavy chain, and topomyosin 1 respectively (Table 3), all key genes during muscle contraction [86]. Similar patterns were found in the reproductive tract of female D. melanogaster, in which muscle contraction genes were up-regulated in response to mating [38]. Such consistent results across flies, honeybees, and moths suggest a conserved function of the female reproductive tract for muscle contraction across insect taxa. Indeed, even in taxa that lack male spermatophores, muscle contraction in the female reproductive tract has been shown to be important in moving sperm into storage and for processing male ejaculates [45].

In contrast, we found little evidence supporting a secretory role for the bursa copulatrix in ECB moths, at least for virgin females. Many of the genes that were up-regulated in the bursa copulatrix compared to the female thorax were not putatively secreted proteins (2%), consistent with the notion that the bursa copulatrix lacks an important secretory function. Nevertheless, we did find that 7% of up-regulated transcripts possessed transmembrane motifs (Table 2), suggesting the presence of receptors or membrane channels that could interact with male-derived proteins.

Bursal gland

Prior to this study, the function of the conspicuous bursal gland present in many lepidopteran female reproductive tracts was completely unknown [49]. Given the direct connection between the bursa copulatrix and the bursal gland, we hypothesized that male-derived products could interact with the bursal gland in two possible ways: by female gland secretions moving into the bursa copulatrix, or by male ejaculate moving into the bursal gland from the bursa copulatrix. Both mechanisms are supported by our transcriptome results, with 16% of up-regulated bursal gland transcripts having secretion signal peptides and 46% having transmembrane motifs (Table 2). Furthermore, unlike the bursa copulatrix the bursal gland had fewer enriched functional categories related to muscle contraction when compared to female thorax (Additional file 4: Table S1), again suggesting mechanical spermatophore breakdown by the bursa copulatrix.

Gene expression in the bursal gland was statistically significantly enriched for many gene ontology categories related to reproduction, including sexual reproduction, gamete generation, multicellular organism reproduction, cellular process involved in reproduction, and developmental processes involved in reproduction when compared to the female thorax (Additional file 4: Table S1). Of these, one ECB female putative protein stands out. This transcript of interest showed homology to purity essence, which has been shown to be involved in sperm individualization and male fertility [87]. Although its specific role in females is unknown, finding this product in a female reproductive tissue suggests that it also plays a role in female reproduction or fertility.

Sexual selection & reproductive isolation

Sexual conflict may extend beyond spermatophore breakdown, with this reproductive “arms race” continuing as males and females struggle for control over fertilization [88]. Such antagonistic sexual coevolution has the potential to contribute to divergence between closely related populations [8, 89]. Proteases and protease inhibitors are two classes of proteins that have been shown to be under positive selection in male reproductive tracts in Drosophila sp. [82, 90]. Although relatively few studies of female reproductive genes have been conducted thus far, these also suggest that proteases and protease inhibitors are important in male–female molecular interactions [42, 43, 83, 91]. Proteases found in the male ejaculate and female reproductive tract have been predicted to co-regulate expression through activation or inhibition of proteolysis, or limit the time for which a reproductive gene or protein is able to act [24, 82]. For example, in D. arizonae 12 digestive proteases were specifically expressed in the female reproductive tract and demonstrated signs of positive selection [92]. The functional role of these proteases is unknown; however the adaptive evolution of digestive proteases in D. arizonae indicates that they likely play a role in male–female molecular interactions [92].

In ECB females, 2,254 transcripts were differentially expressed between the Z and E strains of ECB in either the bursa copulatrix and the bursal gland. Within the bursa copulatrix, 86% (6/7) of statistically significantly enriched categories dealt with protease function and 34 transcripts showed significant homology to proteases or protease inhibitors (Figure 3). Seven of the proteases found in the bursa copulatrix were over-expressed in Z-strain females compared to E-strain females (Table 4), as were seven of the proteases found in the bursal gland (Table 5). In E-strain female bursa copulatrix tissues, comp18651_c0_seq1 had increased expression with a log fold change of 10 and showed significant homology to tryptase 5 (Table 4). Tryptase 5 has been shown to decrease male spermatozoa motility in humans and may be involved in fertility [93]. Another interesting protein, pacifastin-related serine protease inhibitor, was also found to be up-regulated in E-strain female bursa copulatrix tissues (Table 4). Pacifastins have been shown to regulate the immune response, reproduction and phase transition in many insects [94]. Some proteins and inhibitors in the pacifastin family have been shown to have species-specificity in locusts, suggesting they could be important in reproductive isolation [95]. Differential expression of these proteases between strains has the potential to help explain the significant reduction in egg-laying that has been documented when Z-strain females mate with E-strain males during cross-strain matings, but further research using mated females is required to make any conclusions regarding this matter. Other proteases that were differentially expressed between ECB strains in both the bursa copulatrix and the bursal gland were serine and serine-like proteases. Serine proteases have been implicated in increased egg laying after mating seen in many organisms [82]. Serine proteases have also been linked to sperm activation and immune response [82]. This is relevant because mating and sperm storage often leads to changes in regulation of immune response in the female reproductive tract [38, 43, 96], thought to protect females against male-derived pathogens. Although we have identified these proteases in virgin females, these differentially expressed serine proteases present in the ECB female reproductive tract will provide a fruitful path for future study of post-copulatory interactions and post-mating, pre-zygotic barriers.

Conclusions

To fully understand post-mating sexual selection and post-mating, pre-zygotic isolation we must examine reproductive transcripts and proteins in taxa with diverse reproductive structures, physiologies, and mating systems. Much has been learned from Drosophila concerning the molecular interplay between the sexes that takes place after mating, yet little is known about how these reproductive interactions contribute to divergence. Using the European corn borer, we examined female gene expression in the first portions of the female reproductive tract that come in contact with the male ejaculate. Thus, the sequences described here provide initial insight into male and female post-mating molecular interactions in a model for speciation. Our results indicate that sexual conflict over spermatophore breakdown and male–female molecular interactions are likely to be important in Lepidoptera. We found that the main role of the bursa copulatrix is like to be as a muscular sac that mechanically processes the male spermatophore, while the bursal gland appears to serve a secretory function, producing proteins that could interact with male reproductive proteins. We also found evidence that differential expression of proteases between recently diverged strains in both tissues may contribute to post-mating, pre-zygotic reproductive isolation. Our findings represent an important first step in understanding male–female interactions and the link between sexual selection and divergence in lepidopterans. In future studies, examining changes in gene expression profiles during spermatophore processing will provide additional insight into post-mating sexual interactions.

Availability of supporting data

The data set supporting the results of this article is available in the NCBI Transcriptome shotgun assembly project under the accession GAVD00000000 and in the NCBI bioprojects under the accession number PRJNA232037.