Honey bee (Apis mellifera L.) queens, as in most other social Hymenoptera, mate early in life and then store sperm to be used throughout their lifetimes. There has been long-standing interest in the mechanisms that allow social insect sperm to survive at ambient temperatures for years and even decades (Taber & Blum, 1960; Boomsma et al., 2005). For the honey bee, these mechanisms have a practical application with respect to the long-term preservation of semen for artificial insemination (AI) (Laidlaw, 1977; Harbo, 1985; Moritz, 1989). Given restrictions on the movement of honey bee genetic stocks, mandates to select bees resistant to various parasites and pathogens, and new tools available for defining genetic traits, germplasm preservation is a critical requirement for applied work in honey bees.

Our results from antioxidant enzyme activity (Weirich et al., 2002) and gene expression (Collins et al., 2004) assays and the remarkable durability of semen stored alone at room temperature (Collins, 2000b) indicated that the honey bee males (drones) contribute significantly to the physiology of sperm storage. Therefore we initiated the study reported here to identify male-produced proteins involved in honey bee mating and sperm storage.

Extensive studies for Drosophila and several other insects show that the products of male accessory glands have significant effects on female physiology and behaviour (reviewed by Chapman, 2001; Wolfner, 2002; Gillott, 2003; Kubli, 2003), as well as on successful sperm transport and storage (Neubaum & Wolfner, 1999). Eighty-three accessory gland proteins (Acps) have been identified from flies (Wolfner, 2002) some of which have been confirmed to stimulate egg production and laying, reduce female receptivity to mates, mediate sperm storage, and provide microbial protection in the reproductive tract. Both seminal fluid and sperm must be present for the full range of effects to be expressed. We carried out computational searches for honey bee orthologs to the described fly Acps in this project with poor success, suggesting that the known high rates of sequence divergence for many of these proteins might make them indistinguishable across the 300 million years (my) separating flies and bees. Accordingly, we then carried out de novo searches for honey bee proteins implicated in sperm storage, by proteomic (MALDI-TOF peptide mass fingerprinting) analyses of proteins present in the seminal vesicles and in semen itself.

By the time that adult drones emerge from pupation spermatogenesis and spermiogenesis are complete (Bishop, 1920; Hoage & Kessel, 1968). During the first week of adult life, the sperm migrate from the testes to the seminal vesicles (Snodgrass, 1963) where they undergo the final stages of maturation. Figure 1 shows the dramatic difference in size of the testes of immature (Fig. 1a) and mature (Fig. 1b) drones. Also note that the seminal vesicles and the mucus glands become filled in the mature drone. The epithelial cells lining the seminal vesicles secrete a small amount of seminal fluid during the maturation, making up about half of the total semen volume (Verma & Shuel, 1973). At the same time mucus is being produced and stored in the mucus glands.

At the time of mating, muscles in the abdomen of the drone contract creating pressure that everts the genitalia into the queen, turning the reproductive system inside out (Koeniger, 1986). The muscular contractions of the seminal vesicles aid in the ejaculation of the semen into the queen's vaginal passage and on to the median and lateral oviducts. The products of the male mucus gland (Colonello & Hartfelder, 2003) follow the semen and harden on contact to air (Bishop, 1920; AC and VW, personal observation). Immediately after eversion, the male is paralysed and falls away from the queen. The mucus plus the endophallus and cornua portions of the male genitalia, which are left behind, become the mating sign, visible outside of the queen's abdomen. From comparisons of mating behaviour in several species of Apis (Koeniger, 1990; Koeniger & Koeniger, 2000) it appears that the mucus and genital structures hold the drone in place during sperm transfer and the mating sign becomes a signal to other males following the queen. Koeniger (1986) proposed that the mucus may also serve to hold the queen's sting out of the way of the next mate.

The honey bee queen makes one to four flights (Roberts, 1944) away from the colony when she is about a week old, mating with from seven to 44 males in rapid succession (Taber, 1954; Moritz et al., 1996; Neumann et al., 1999; Tarpy & Page, 2001). Over the next 24–48 h, the sperm migrate to the spermatheca (Ruttner et al., 1971). This movement is enabled by contractions in the queen's abdomen (Koeniger, 1986), by sperm motility (Collins, 2000a) and the presence of spermathecal fluid (Gessner & Ruttner, 1977). Only about 3–5% of the sperm are actually retained in the spermatheca (Koeniger & Koeniger, 2000), the rest are lost from the queen.

Genetic studies using phenotypic markers have shown that sperm from all of the mated drones become randomly distributed within the spermatheca (Page et al., 1984) and relatively constant levels of all paternal types are represented in the worker offspring. Therefore there is little or no sperm competition, although different drones may contribute variable numbers of viable sperm (Woyke & Jasinski, 1978; Collins, 2004). The creation of colonies of multiple paternal lines of workers, or subfamilies, is evolutionarily desirable, as these colonies have the capability of responding readily to wide changes in the environment (Jones et al., 2004) and extreme polyandry increases the fitness of the queen by reducing the colony-level impact of her laying non-viable, diploid drone eggs (Tarpy & Page, 2001).

A total of 234 spots were cut from 2-dimensional gels and analysed using peptide mass fingerprinting (MALDI-TOF) against the Genbank nunredundant (nr) and official bee protein sets, 90 from seminal vesicle tissue and 144 from semen, 90 picked from one gel (pH 3–10) and 54 from a second gel (pH 7–10). Using Mascot and the bee-specific database, we found significant GLEAN3 matches from 60 protein spots derived from seminal vesicles (54 unique protein IDs), plus one additional match to a novel peptide predicted by Genscan. We found 78 GLEAN3 matches to genes present in honey bee semen (33 of these are unique). Because we ran gels with overlapping pH ranges for the separation of semen proteins, there was greater redundancy of spots from these samples. There was also some biological redundancy with several neighbouring spots showing the same database matches. Tissues and semen were collected from six source drones and it is possible that redundant spots reflect both allelic (isozymic) variation as well as variation in the production or post-translational processing of proteins. In support of the latter, many identical matches were derived from spots with significantly different size or pH traits (Table 3). Interestingly we found no spots that matched non-bee entries in the GenBank database that did not also match a honey bee gene, attesting to both the completeness of the bee gene list and the difficulty in making Mascot fingerprinting identifications over this long evolutionary distance. A complete listing of all spots and results of analyses are listed in the Supplementary table.

Of the 69 unique honey bee proteins found 66 showed significant BLASTP matches to proteins in the genome of Drosophila melanogaster (e-value < 1.0 × e⁻¹⁰, Tables 1 and 2). Two of the remaining proteins (GB11987, spot A13; GB13059, spots A15, B85, C3; and GB18752, spots B90, C44) showed no non-bee matches in nr while the third de novo predicted gene (Group 1.17.1, spot A47) seems to be a widespread protein lost from the fly genome.

Eighteen identified proteins were found in both seminal vesicle and semen samples, 36 were unique to the seminal vesicle and only 15 were unique to semen (although they might have come from somewhere else in the male reproductive tract). This latter number is an under-representation of semen-unique proteins by Fisher's exact test (P = 0.04) (Sokal & Rohlf, 1995). Proteins were grouped by Gene Ontology biological processes using the parent terms cytoskeleton, defense, energy, glycolysis, glyoxylate cycle, metabolism, muscle, tricarboxylic acid cycle or other (Tables 1 and 2). The identified proteins from this sampling were diverse (37 of 69 had distinct GO biological functions) but included some apparent functional biases. Overall, the proteins showed an abundance of genes involved in metabolism and cytoskeletal function. The GO parent groupings fell out differentially for the two protein sources (Table 4; global G-test, P = 0.003), with an excess of muscle-related and cytoskeletal genes in the seminal-vesicle proteins and an excess of proteins related to glycolysis (and possibly the glyoxylate cycle) in the semen samples. Most proteins (38 of 66 unique GB matches) were predicted by pSortII to reside in the cytoplasm, followed by mitochondria (18 of 66). Only two proteins (GB12716 and GB18752) were predicted to be extracellular, and both of these were found only in the semen component.

Sperm is stored by honey bee queens, and queens of other social insects, for years prior to fertilization. In honey bees, sperm has also been shown to remain viable for extended periods in vitro at ambient temperatures. Collins (2000a) held semen alone in sealed capillary tubes up to 1 year, and found viability did not begin to drop until about 9 weeks and in some samples remained above 50% for up to 6 months, a remarkable survival rate compared with other higher animals.

This survey provides the first global look at sperm-related proteins in honey bees and the first, to our knowledge, genome-enabled proteomic survey of semen proteins in an insect. The results indicate constitutional biases in sperm-related proteins as well as information on the potential sources of semen proteins. A high number of the proteins identified in these samples showed significant matches to Drosophila (66 of 69 at < e⁻¹⁰) indicating that this is a fairly conserved set of proteins.

When organized into GO groups there is an apparent over-representation of genes implicated in the glycolysis pathway (10 of 55 proteins that could be associated with a GO biological function). Other carbohydrate metabolism genes were also present (from the Krebs’ and glyoxylate cycles) along with two antioxidative enzymes, superoxide dismutase (SOD), and glutathione-S-transferase (GOT). Regarding SOD and GOT, previous studies of enzyme activity (Weirich et al., 2002) and gene expression (Collins et al., 2004) targeted these as two of three potential antioxidants in reproductive tissues. SOD was expressed in semen, one-third of the activity directly associated with the sperm fraction, and at similar levels in the spermathecae of mated and virgin queens. GOT was found in the spermatheca at higher levels after mating, but not in semen. Our third subject enzyme in these previous studies, catalase, was significant in that it was almost entirely associated with the sperm itself, as opposed to seminal fluid and was also expressed at higher levels in spermathecae of mated queens (Weirich et al., 2002). It is surprising that we did not identify catalase in the current survey, a result perhaps reflective of its low abundance in seminal fluid and/or its membrane association in sperm.

We found significant biases in metabolism-associated proteins in semen vs. the seminal vesicle. Blum et al. (1962) found high levels of sugars, especially glucose, fructose and trehalose, in the seminal fluid as well as phospholipids. The sugars were rapidly oxidized. Verma & Shuel (1973; Verma, 1978) speculated that the sugars and phospholipids were the primary sources of energy for motility of sperm. Previous observations of motility in honey bee semen by Poole & Edwards (1970) demonstrated that the spermatozoa became non-motile within a short period, but the addition of sugars to the semen restored motility. Verma & Shuel, (1973) also found that sperm were motile for only 6 h in an extender with no oxidizable substrate. While we can not directly infer the metabolic processes endogenous to sperm (and seminal fluid) from this project, the protein biases are suggestive of the specific metabolic (and catabolic) activities that sustain sperm prior to and during mating.

A comparative study of mammalian, avian and honey bee semen by Kraft et al. (1978) reported that honey bee sperm were unusual in that they used only the Embden–Meyerhof glycolysis cycle for metabolism, and not the Krebs’ cycle. They commented that this would be highly adaptive for sperm that remained inactive for extended periods. We found no TCA enzymes in semen, although there were two in the seminal vesicles.

The Acps from flies rarely have orthologs in bees based on the honey bee genome annotation (HBGSC, 2006). This may reflect shorter Acp length, but it seems more likely that this is a corroboration of the dogma of fast-evolution for these genes. Mueller et al. (2004) used comparative structural modelling to infer conserved protein functional classes of 52 of the predicted seminal proteins. These classes included: regulators of proteolysis, lipid-modifying enzymes (lipase-related), lectins, cysteine-rich proteins of the CRISP family, an antioxidant, antimicrobial peptides, and one RNase. In our continuing studies of the reproductive tissues of the honey bee, we will look for proteins of similar function. The proteolytic proteins may be involved in sperm motility, and similar Acps in Drosophila may serve as coagulation factors related to mating plug formation. Proteins with lipid-binding sites and lectins would participate in lipid and carbohydrate metabolism, respectively, already suggested as the energy sources for honey bee sperm (Verma & Shuel, 1973). The presence of an antioxidant agrees with our own studies showing three such enzymes as important components of semen and present at high levels in the spermatheca. We did not include mucus glands in the present study, but Colonello & Hartfelder (2003) reported only three dominant and three minor polypeptides present.

High rates of sequence evolution aside, it is not too surprising that a group of proteins intimately associated with reproductive biology are dissimilar between two insects with such different reproductive behaviour and physiology. The Drosophila female generally mates with one male at a time, although she may mate again after oviposition. There is behavioural competition between male Drosophila for mates during extensive courtship rituals. The matings generally last for 15–20 min, also excluding other males, and there is documented sperm competition for fertilization (Simmons, 2001). Proteins from the male accessory gland in seminal fluid and the mating plug aid in protection of the male investment by reducing the female tendency to mate again, and spur the initiation of egg laying. For the honey bee drone the key competitive hurdle is in reaching the female first from the drone ‘comet’ that follows her in flight – a competition that relies more on visual acuity and flight physiology. Matings are rapid, taking a matter of seconds, and females often mate several times in immediate succession (Koeniger, 1986). In contrast to many insects (Gillott, 2003) the honey bee mating plug does not seem to negatively affect the success of subsequent mates, and indeed might actually aid this success (Koeniger, 1986). Nor does it affect the future viability of the queen (Colonello & Hartfelder, 2005). While mating barriers, per se, seem less likely in honey bees than in other insects, the possibility still remains that the drone-provided proteins have strong effects on the physiology of queens.

Drosophila males have one pair of accessory glands with two cell types (Chapman & Davis, 2004); drones have two separate glands, paired seminal vesicles and mucus glands. Similar research on the mucus gland of the honey bee shows that by maturity the number of components has been reduced to three dominant polypeptides and three minor ones (Colonello & Hartfelder, 2003). For the bumble bee, Bombus terrestris, Baer et al. (2000, 2001) report that the mucus plug contains four fatty acids and a dipeptide, but it is the fatty acid, linoleic acid, which inhibits remating. This species mates only once. The mucus/genitalia-tissue plug of the stingless bee, Melipona quadrifasciata, inhibits remating also, but simply by the mechanical stimulation of the queen's abdomen (Melo et al., 2001).

We know little or nothing about changes that take place in the spermatozoa as they move from the testes to the seminal vesicles to the spermatheca to fertilization of the egg. In other domestic animals, primarily mammals, there are significant changes (activation, capacitation and the acrosome reaction) that occur during the migration within the male and after deposition in the female reproductive tract that are critical to the final act of fertilization of the egg. However, there have been few indications of similar shifts in physiology during sperm maturation or sperm storage in insects. The proteins found in the seminal vesicle and semen might guide us to a better understanding of sperm developmental stages in the honey bee and thus to insects in general.

Similar proteomic and genomic-centred analyses continue with a focus on the spermatheca and its contents, before and after mating. Differences in pH between the semen and reproductive tract that function for orientation of the sperm to the spermathecal duct, may be only one factor in the changed metabolic processes during storage. Future studies also need to investigate enzymes in testes, where spermiogenesis occurs, as well as the ejaculatory and spermathecal ducts. In a histochemical study Schoeters & Billen (2000) found polysaccharides in spermathecal ducts of queens but not workers, and suggested that these sugars provided energy for activation of sperm prior to fertilization. Klenk et al. (2004) identified an SDS–PAGE gel band from spermathecal fluid as a 29 kDa protein somewhat homologous to triosephosphate isomerase. They suggested that it was not serving a glycolytic function in the queen but aiding in sperm storage. Results from our continuing studies will provide more insight into the remarkable durability of honey bee sperm and how it is supported by the queen's physiology.

The work reported here highlights the invaluable data that have been made available through the Honey Bee Genome Project. Our initial peptide mass fingerprinting efforts using existing databases for honey bees and other insects showed < 5% hit rates, a 10-fold lower rate than when the honey bee genomic data were included. The availability of a genome-wide predicted bee protein list and the genome sequences themselves was essential to this project and will provide extensive new opportunities for proteomic research in honey bees.