A network of interactions among seminal proteins underlies the long-term postmating response in Drosophila

Abstract

Despite the importance of seminal proteins in fertility and their capacity to alter mated females' physiology, the molecular pathways and networks through which they act have not been well characterized. Drosophila seminal fluid includes proteins that fall into biochemical classes conserved from insects to mammals, making it an excellent model with which to address this question. Drosophila seminal fluid also contains a “sex peptide” (SP, Acp70A) that plays a major role in regulating egg production and mating behavior in females for several days after mating. This long-term postmating response (LTR) initially requires the association of SP with sperm. The LTR also requires members of the conserved seminal protein classes (two lectins, a protease, and a cysteine-rich secretory protein). Here, we show that these seminal proteins function interdependently, regulating a three-step cascade (first, at the level of seminal protein transfer to the female; second, at the level of stability; and third, at the level of localization within females), leading to the normal localization of SP to sperm-storage organs. This localization is, in turn, necessary for successful induction of the LTR. The requirements for manifestation of the LTR in Drosophila establish the paradigm that multiple seminal proteins can exert their actions through a multistep, multicomponent network of interactions.

In sexually reproducing organisms, molecular contributions from both males and females are essential for successful reproduction. Males provide sperm and seminal fluid to the female. These contributions induce physiological and behavioral changes in mated females. Seminal fluid proteins have been identified in a range of animals (reviewed in refs. 1–3), and loss or reduction in levels of some of these molecules can impair fertility (examples reviewed in refs. 1–3). Thus, unraveling their mechanisms of action will offer insights of potential relevance to understanding some causes of human infertility and to designing insect pest management strategies. Yet despite their importance, the molecular actions of seminal proteins, even those in biochemical classes conserved from insects to mammals (4), are poorly understood. An exception is that a few seminal proteases mediate proteolytic pathways [e.g., semenogelin processing in primates (5–7) and ovulin processing in Drosophila (8, 9)]. We show here that Drosophila seminal fluid proteins falling into several conserved biochemical classes interact to regulate the persistence of postmating responses.

In Drosophila, an important subset of seminal proteins is made in the male's accessory glands. These accessory gland proteins (Acps) induce postmating changes in the physiology and behavior of mated females (reviewed in refs. 3 and 10). Long-term persistence of these postmating responses [the long-term postmating response (LTR) (11–15)] requires both Acps and stored sperm in the female. A critical component of the LTR is an Acp called sex peptide [SP, Acp70A (16, 17)]. On its own, SP can induce increased in egg production and can reduce in receptivity to remating, but these effects are transient (18). Upon transfer to females during a normal mating, SP binds to sperm and persists in mated females' sperm-storage organs for ≥5 days postmating (14). Sex peptide's gradual release from stored sperm is suggested to allow it to access its target cells to induce postmating responses over an extended period. Specifically, SP has been shown to regulate egg-laying and receptivity behaviors through the action of its G-protein-coupled receptor in a group of ppk⁺fru⁺ neurons associated with the female reproductive tract (19–21).

Recently, four additional Acps [lectins CG1652 and CG1656, protease CG9997, and cysteine-rich secretory protein (CRISP) CG17575] were shown to also be required for manifestation of the LTR (4, 15). All of these Acps tested were undetectable in female by 24 h postmating (22). This suggests the hypothesis that the four short-lived Acps effect the LTR by allowing SP to be retained within the female, for example, by promoting SP's localization to sperm-storage organs. To test this hypothesis, we analyzed protein levels and localization of each of these Acps in the absence or knockdown of each of the remaining four Acps. We show that CG1652/CG1656, CG9997, and CG17575 interact to mediate the normal localization of SP to sperm and sperm-storage organs.

Results and Discussion

Long-Term Persistence of SP in Mated Females Requires CG1652/CG1656, CG9997, and CG17575.

To test whether CG1652/CG1656, CG9997, or CG17575 are required for the transfer of SP to the female reproductive tract, we compared the accumulation and levels of SP in females mated to males deficient in each of these Acps. As parallel controls, we tested SP persistence in females mated to the normal siblings of the experimental males or to males deficient in CG33943, an Acp that does not affect long-term postmating responses (15). We did not observe detectable differences in SP transfer or levels in females mated to control or any of the knockdown males at 2 h after the start of mating (ASM) (Fig. 1A). That knockdown males transfer normal amounts of SP to females implies that they had made normal amounts of this Acp. From this, we infer that synthesis of SP in males and SP transfer to females are independent of CG9997, CG1652/CG1656, or CG17575.

Fig. 1.

Levels of different accessory gland proteins (Acps) in mates of control or Acp-deficient males. Samples are: reproductive tracts (RTs) (A) or levels of SP in seminal receptacles (SRs) (B) of females mated to CG1652, CG1656, CG9997, CG17575, or CG33943 control (+) or knockdown (−) males at 2 h after the start of mating (ASM). Female RTs (minus ovaries, here and in remaining blots) or SRs were dissected at 2 h ASM. Protein equivalents of four mated female RTs or 10 SRs were loaded in each lane. The blot was probed with anti-sex peptide (SP row) or with CG1652, CG1656, CG9997, or CG17575 antibodies, as indicated in successive rows. (C) Levels of SP in SRs at 24 h ASM. Although we represented CG1652 and CG1656 independently, knockdown of either one knocks down both (ref. 15; see Materials and Methods for details), because CG1652 and CG1656 are gene duplicates (32). The AG lane contained protein equivalents of one pair of accessory glands from Canton-S males in this blot (and the remaining figures in this article). The blots were probed with anti-α-tubulin as a loading control (tubulin row). The AG samples typically show low signal with tubulin antibody. CG9997 is processed into a smaller form in the mated female; arrows indicate the full-length 45-kDa and processed 36-kDa forms in this figure (and in the remaining figures in this article). All of the samples were resolved on 7.5–15% gradient SDS/PAGE in this figure (and in the remaining Western blots in this article).

Upon transfer to females, SP binds to sperm and persists in mated females' seminal receptacles (SRs) for ≥5 days postmating (14). We tested whether the four short-lived LTR Acps affect the accumulation of SP in SRs (see Materials and Methods for details) of mated females. Although SP normally persists ≥5 days in SRs (14, 17), we observed less or no SP in SRs of females mated to CG1652/CG1656, CG9997, or CG17575 knockdown males (as compared with controls) even at 2 h ASM (Fig. 1B). By 24 h ASM, we no longer detect SP in SRs of CG1652/CG1656, CG9997, or CG17575 knockdown mates, in contrast with controls or parallel controls (CG33943) (Fig. 1C). Reduced SP localization to SRs in the absence of each of these four Acps does not reflect a lack of sperm in storage, because mates of knockdown males contain sperm in their SRs at levels similar to or higher than those of their controls (15). Therefore, CG1652/CG1656, CG9997, and CG17575 are required for localization or retention of SP in the SR. In contrast, SP is not required for the transfer of CG1652, CG1656, CG9997, and CG17575 to females and/or for subsequent accumulation of the former three to SRs (SI Text and Fig. S1).

Sex peptide binds to sperm and accumulates in SRs (11, 14). Sperm are necessary for SP accumulation in the SR (SI Text and Fig. S2). Because very little SP accumulates in SRs in the absence of CG1652/CG1656, CG9997, or CG17575 (Fig. 1B), we tested whether SP binding to sperm depends on the presence of these Acps. For this purpose, we dissected sperm out of the SRs of females mated to CG1652/CG1656, CG9997, and CG17575 control or knockdown males at 2 h ASM and probed both the sperm sample and the remaining empty SR samples (see caption to Fig. 2 for details) with anti-SP antibody. We detected sperm-bound SP in controls (as expected from ref. 14). Interestingly, we detected high levels of SP in SRs from controls, even after we had removed sperm by dissection [Fig. 2, controls (+), also control (+) and knockdown (−) in CG33943 row], suggesting that SP also binds to SRs. In contrast, we found only traces of SP in samples of sperm or of SRs lacking sperm from females mated to CG1652/CG1656, CG9997, or CG17575 knockdown males [Fig. 2, knockdown (−)].

Fig. 2.

Localization of sex peptide (SP) within seminal receptacles (SRs) of females mated to males knocked down (lane, −) for CG1652/CG1656, CG9997, CG17575, or CG33943 or mated to their control males (lane, +). Seminal receptacles from mated female reproductive tracts were dissected at 1 h after the start of mating (ASM), and sperm were pulled out after puncturing the middle of the SRs. Protein was extracted from sperm (lane, sperm) or SRs minus sperm [lane, SR (no sperm)] and processed for Western blot analysis with anti-SP (or anti-α-tubulin as a loading control). To confirm that SR (no sperm) samples contain no or few sperm, sperm were counted (see Materials and Methods for details) in those SRs after pulling out the sperm; we found only ≈25 sperm in those SRs (n = 5), whereas normally SRs contain 250–300 sperm at this time point. Each lane contains protein equivalents of 20 females' stored sperm or SRs lacking sperm.

These results indicate that in normal matings SP binds to sperm as well as to the SR and that such association requires CG1652/CG1656, CG9997, and CG17575. Our finding that association of SP with sperm requires these other Acps explains why recapitulating the normal binding of SP (i.e., specific to sperm tails) by incubating pure SP with sperm dissected from males' seminal vesicles has not been possible (23); the additional Acps needed to mediate the specific binding were not present in those incubations. The need for several Acps to mediate SP association with sperm also suggests the intriguing possibility that these Acps' action might act as a sort of timer controlling SP's localization. If SP entered the female unbound to sperm, then it therefore initially would be free to cross a permeable part of the female's reproductive tract to enter the female's hemolymph and access targets such as the corpora allata, where it could modulate juvenile hormone synthesis (24–26). However, later, once the actions of the Acps described here caused the SP to associate with sperm, SP would be present in the sperm-storage organs, from where it could be slowly released to continue maintaining postmating responses.

CG9997 Is Required for the Normal Transfer of CG1652 and CG1656.

Because we observed similar effects on SP upon knockdown of CG1652/CG1656, CG9997, or CG17575, we hypothesized that these Acps could interact in a network. To test this, we first analyzed the requirement of each Acp for the transfer of the others to females by comparing females mated to Acp-deficient males to their controls. CG9997, CG1656, and CG1652 do not require CG17575 for synthesis or transfer, because their levels did not differ between mates of CG17575 knockdown and control males and vice versa (Fig. 1A). In contrast, females mated to CG9997 knockdown males had greatly reduced levels of CG1652 and CG1656 in their uteri compared with those of controls at 2 h ASM [Fig. 1A, control (+) and RNAi (−) lanes of CG1652 and CG1656 in 9997 column]. These females received other Acps tested at levels similar to controls (ref. 15; Fig. 1A, SP and CG17575 rows). Thus, CG9997 affects either the transfer of CG1652 and CG1656 or their retention in mated females. Subsequent analysis of reproductive tract samples from CG9997 control and knockdown mates at 10 min and 1 h ASM indicated that CG9997 is required for the normal transfer of CG1652 and CG1656: Females mated to CG9997 knockdown males had less CG1652 and CG1656, even at early times during mating (Fig. 3 and see SI Text).

Fig. 3.

CG1652 and CG1656 in females mated to CG9997 control (+) or knockdown (−) males at 10 min and 1 h after the start of mating (ASM). Protein equivalents of four reproductive tracts (RTs) were loaded into each lane and processed for Western blot analysis with anti-CG1652 (CG1652 row), and CG1656 (CG1656 row), CG9997 (negative control), Acp29AB (positive control for the normal transfer of other Acps/lectins), and α-tubulin (loading control) antibodies as indicated.

CG1652, CG1656, or Both Are Required for the Stability of CG9997 in Mated Females.

To further define the interactions among these Acps, we tested whether CG9997 requires the other Acps for transfer and tissue targeting. CG9997 is normally detected as a 45-kDa protein in males and undergoes processing in the male during transit to the female at the time of mating. Both full length (45 kDa) and processed (36 kDa) CG9997 are detected in mated females' reproductive tracts (see Fig. 1A, Fig. S3, and SI Text for details). Females mated to CG17575 control or knockdown males had 45- and 36-kDa products of CG9997 at 2 h ASM (Fig. 1A, CG9997 in 17575 column), showing that transfer and processing of CG9997 are independent of CG17575. In contrast, females mated to CG1652/CG1656 knockdown males contained relatively lower levels of CG9997 than controls and had only the 36-kDa form of CG9997 [Fig. 1A, CG9997 in RNAi (−) lanes of 1652 and 1656 columns]. A time course analysis of CG9997 on samples collected from females during and after mating to CG1652/CG1656 knockdown or control males showed that females mated to CG1652/CG1656 knockdown males had the 36-kDa band but little to no 45-kDa band of CG9997 from 15 min to 2 h ASM (Fig. S4) unlike their controls, which had both (Fig. S4). This suggests that CG1652, CG1656, or both are required for the molecular stability of CG9997; whether effects of CG1652/CG1656 on stability and on processing of CG9997 are independent is not presently clear. Nevertheless, it suggests that CG9997 (protease) and CG1652/CG1656 (lectins) function interdependently and also the possibility of regulatory interactions between seminal proteases and seminal lectins in reproduction.

CG17575 Is Required for CG1652 and CG1656 to Localize to the Seminal Receptacle.

Above, we reported that CG1652/CG1656, CG9997, and CG17575 are required for the accumulation of SP in the SRs of mated females. We wished to determine whether any Acp in this group was required for the accumulation of any other Acp(s) in this group within the SR. Because of technical limitations in detection of some Acps (see SI Text), we could only analyze the requirement for CG17575 in the accumulation of CG1652, CG1656, and CG9997 in the mated female's SR.

We detected CG9997 in SRs of females mated to CG17575 knockdown males at levels similar to those of controls (Fig. 4, SR lanes in CG9997 row), suggesting that CG17575 is not necessary for transfer, stability, and accumulation of CG9997 and vice versa (also see Fig. 1A). In contrast, we detected CG1652 and CG1656 only in SRs of females mated to CG17575 control males [Fig. 4, SR control (+) lanes in CG1652 and CG1656 rows] but not in those of mates of CG17575 knockdown males [Fig. 4, SR RNAi (−) lanes in CG1652 and CG1656 rows]. Sperm are required for the accumulation of CG1652, CG1656, and CG9997 in sperm-storage organs (SI Text and Fig. S2). However, differences in levels of CG1652 and CG1656 in SRs of mates of CG17575 knockdown versus control males do not reflect differences in sperm storage given the normal sperm counts at 2 h ASM (15) and at 1 h ASM (control mates, 486 ± 12.46; knockdown mates, 514 ± 30.59; P = 0.43). Rather, CG1652 and CG1656 require CG17575 for their localization to SRs.

Fig. 4.

Transfer and localization of CG1652, CG1656, and CG9997 to seminal receptacles (SRs) in females mated to CG17575 control or knockdown males. Samples prepared from SRs dissected from reproductive tracts and the remaining reproductive tracts (RTs) at 1 h after the start of mating were processed for Western blot analysis and probed with anti-CG1656, CG1652, CG9997, and α-tubulin antibodies. Protein equivalents of 20 SRs and 10 female RTs were loaded.

Above, we showed that CG9997 is required for transfer of CG1652 and CG1656 to females. Here, we show that CG17575 controls accumulation of the same two Acps in SRs. Given these effects, the lack of SP localization to sperm-storage organs in females mated to CG9997 or CG17575 knockdown males possibly reflects a role for CG1652 and CG1656 and that CG9997 and CG17575's effects on SP localization are thus indirect (through CG1652/CG1656). However, we cannot rule out at present a direct effect of CG9997 and CG17575 on the accumulation of SP to sperm-storage organs.

CG1656 Binds to Sperm, and This Binding Requires CG17575.

We showed, above, that CG17575 is required for the accumulation of SP, CG1652, and CG1656 in the SR and for the normal association of SP with sperm (Fig. 2). However, we did not detect CG17575 in mated females' SRs upon transfer (ref. 22; Fig. S1 and Fig. S2). These observations suggested that for CG17575 to exert its effects, it and the other LTR-promoting Acps might interact with one another, with other male- or female-derived molecules, or with both, before entry of sperm and Acps into SRs, for example, when Acps and sperm are still in the uteri of mated females (refs. 14, 17, and 22; Figs. S1 and S2). Because SP association with sperm requires CG17575, we wished to explore whether the LTR-promoting Acps CG1652 and CG1656 also associated with stored sperm and whether this association requires CG17575. We tested this by immunofluorescence on sperm dissected out of SRs at 1 h ASM. Due to the cross-reactivity of anti-CG1652 with other proteins, we could only perform these experiments with anti-CG1656. We detected CG1656 on the tails, but not heads, of sperm from SRs of mated females in all controls (including knockdown of CG33943, which does not affect the LTR) or in mates of SP knockout males (no effect on CG1656 localization, see above) (Fig. 5A, C–G, and I). In contrast, we did not detect CG1656 on sperm from females mated to CG1656, CG9997, or CG17575 knockdown males (Fig. 5 B, H, and J). These results indicate that CG1656 associates with sperm tails and that this association requires CG17575. The lack of detectable CG1656 on sperm tails from CG9997 knockdown males is likely a consequence of the requirement for CG9997 for the normal transfer of CG1656 into the female (see above, Fig. 3). At present, whether CG17575 plays a direct or indirect role in associating CG1656 with sperm (or in allowing it to enter the SR) cannot be determined. Thus, we find that a Drosophila seminal CRISP (CG17575) plays an important role in mediating the association of a lectin (CG1656) with sperm and with a sperm-storage organ. This is also interesting given that CRISPs and lectins in mammalian seminal fluid are suggested to be involved in sperm function (reviewed in refs. 27 and 28).

Fig. 5.

Immunofluorescence detection of CG1656 on sperm from seminal receptacles (SRs) of females mated to control (Left) or knockdown (Right) males at 1 h after the start of mating. (A) CG1656 is a control for the specificity of the antibody. In these overlay images, anti-CG1656 staining (green) is seen on sperm tails [but not on sperm heads (stained red with propidium iodide)] in the control mating and is undetectable in mates of CG1656 knockdown males (B), as expected because they do not receive CG1656. Anti-CG1656 staining of sperm from SRs of females mated to control males (C, E, G, I) or knockdown males (D, F, H, J) lacking the indicated Acps also is shown. Staining was as in (A, B).

Model for Acp Interactions That Establish the Long-Term Response.

On the basis of our results (summarized in Table S1), we propose the existence of a regulatory “LTR network” of Acps (Fig. 6) that mediates the localization of SP to sperm and SR. That SP then can be gradually released from those sperm to access targets that will allow it to mediate the LTR (14). Importantly, 138 seminal proteins have been identified so far in Drosophila melanogaster (reviewed in refs. 1 and 3; see also ref. 29). Thus, future functional analyses of additional seminal proteins are likely to add more players, and perhaps more steps, to this LTR network.

Fig. 6.

A model for the network of interactions among accessory gland proteins (Acps) and events that set up the long-term postmating response (LTR) in mated Drosophila females. Highlighting denotes Acps that function at those particular steps in the LTR interaction network; question marks denote uncertainties, such as whether action of CG17575 on CG1656 localization is direct or whether or how sex peptide (SP) is released from its seminal receptacle (SR) binding. CG9997 assists in the transfer of CG1652 and CG1656 (CG17575 and SP are transferred without its assistance). The protease CG9997 appears to be the most upstream member in the LTR network (thus far). (Transfer of CG9997 does not depend on any other Acps in the network.) After entry into the female, the lectins CG1652 and CG1656 regulate stability, processing, or both of CG9997. In the female's uterus, the cysteine-rich secretory protein CG17575 action allows, or is a precondition for, the binding of CG1656 to the sperm and accumulation of CG1652 and CG1656 in SRs. Sex peptide appears to be the most downstream molecule in the LTR network, because it requires CG1652/CG1656, CG9997, and CG17575 for its accumulation in the SR. Gradual release of SP from sperm tails and SRs is proposed to allow it to access its targets and cause the LTR (17–19).

Most Acps in the LTR network are in conserved protein families found in seminal fluid; our results therefore provide a model for dissecting seminal fluid regulatory cascades in other animals. Our results also have implications for evolutionary biology. Some seminal proteins evolve rapidly (30–35), which has been suggested to be the consequence of sexually antagonistic coevolution (36) or intrasexual competition. That some seminal proteins function interdependently as shown here could constrain their ability to evolve rapidly or might imply that only certain members of the network are in a position to do so. Consistent with the former notion, CG17575 is highly conserved (31, 32), whereas CG9997 shows evidence for rapid evolution (31, 35). Further, our results also suggest that LTR candidates listed here, particularly the rapidly evolving ones, may be useful targets for genetic association studies in Drosophila (37). Despite the conservation of SP and its receptor across Drosophila [and beyond in the case of the receptor (21)], outbred males show genetic variation in their defensive sperm competitive abilities and in their ability to cause female refractoriness to remating (38, 39). Therefore, studying whether these phenotypic differences are attributable to LTR candidates mediating SP's functions in mated females will be interesting.

To conclude, we have shown that interdependent interactions among a group of seminal proteins result in the localization of a long-term mediator, the SP, within the sperm-storage organs, thus leading to the manifestation of the LTR. Previous work in Drosophila and in primates has identified specific enzymatic effects of seminal proteins on one another (5–8). Here, using Drosophila as a model because of its tractable genetics, we show that interactions among seminal proteins are even more multifaceted: Seminal proteins work in multistep interactive networks that go beyond simple enzymatic interactions to regulate each others' localization, sperm association, and stability or transfer in females. Given that these molecules fall into conserved seminal fluid protein classes (4), we believe that the sort of interactions defined in the present study are also likely to be a model for seminal protein action in animals beyond Drosophila. Similar to the requirement of females' environment for proteolytic cleavage of some seminal proteins (9, 40), opportunities might exist for female modulation of, and participation in, seminal protein interactions such as the ones we report here. Dissection of such interaction networks is also important for consideration in strategies to diagnose or alleviate certain infertilities, to control the reproduction of insect pests, or both.

Materials and Methods

Flies.

We used transgenic fly lines carrying sympUAST-Acp (UAS-Acp-UAS) constructs for CG1652/CG1656, CG9997, and CG17575 to knock down mRNA and protein levels of these Acps (9, 15). The nonknockdown siblings of these flies were controls. As an additional control for the strain background, we carried out parallel crosses and experiments with flies carrying a sympUAST construct for knockdown of CG33943, which is an Acp that does not affect LTR (15). As described in refs. 9 and 15, we crossed the sympUAST-Acp lines to tubulin-GAL4/TM3, Sb flies (41) to generate the knockdown (experimental) males (tubulin-GAL4;UAS-Acp-UAS) and control males (TM3, Sb/UAS-Acp-UAS). Two independent insertion lines were analyzed for each Acp knockdown. The RNAi lines tested here knocked down the targeted Acp to <2.5% of the wild-type levels (9, 15). CG1652 and CG1656 are gene duplicates (4, 32) with ≈80% identity at the nucleotide level; RNAi of either one knocks down both (15). Hence, we cannot at present determine whether the observed phenotypes with their knockdown lines are due to both of these genes or to one in particular. Therefore, for the ease of discussion pertaining to results from these knockdown lines, we represent them as CG1652/CG1656 (also see the note below in Sample Preparation and Western Blot Analysis section). For SP, we used an SP null mutant (17) kindly provided by Eric Kubli (University of Zürich, Switzerland). Sex peptide null mutant males and their sibling control males were generated by crossing the SP knockout line (0325/TM3, Sb ry) to a deficiency line (Δ130/TM3, Sb ry), as described in refs. 14 and 17.

Matings were carried out by crossing 3- to 5-day-old unmated SP mutant or Acp knockdown or control males to 3- to 5-day-old virgin females of the Canton-S strain of D. melanogaster. To test for the requirement of sperm for the entry of Acps into sperm-storage organs, spermless males were generated by crossing tudor (tud¹ bw sp/tud¹ bw sp; ref. 42) females to Canton-S males. Isogenic control males for this experiment were the Cy⁺ sons of tud¹ bw sp/CyO females crossed with Canton-S males; we verified that these flies made sperm. All flies were maintained on standard yeast–glucose medium at room temperature (22 ± 1 °C) and 12:12 light/dark cycle.

Sample Preparation and Western Blot Analysis.

To determine the efficiency of transfer of Acps, we used Western blots to do a time course analysis using protein samples prepared from reproductive tracts of mated females as in Ravi Ram et al. (22). Three- to 5-day-old Canton-S virgin females were mated to 3- to 5-day-old unmated control or knockdown males or SP null males. The mating pairs (mating normally lasts for ≈18–25 min in these flies) were frozen at 15 min ASM, or matings were allowed to complete without interruption, and mated females were frozen at 30 min, 1, and 2 h ASM.

Localization of Acps to SRs in the absence of CG1652/CG1656, CG9997, CG17575, or SP was analyzed using protein samples containing 18–20 SRs dissected from the mates of control, Acp knockdown, or SP null males at 1 h ASM. The localization of SP to the SRs in the absence of CG1652/CG1656, CG9997, or CG17575 was analyzed at 2, 24 h, and 4 days ASM. Although Drosophila has two types of sperm-storage organs (SR and spermathecae), we analyzed only the samples from SRs, for three reasons. First, females mated to males knocked down for CG1652/CG1656, CG9997, or CG17575 had significantly more sperm in their SRs compared with controls, but levels of sperm in their spermathecae were similar to those of controls (15). Second, sperm are easy to retrieve from SRs but difficult to retrieve intact from spermathecae during dissections for sample preparation. Third, only SRs had been analyzed in the previous study that showed SP binding to the sperm (14).

Sample preparation and Western blot analysis were carried out as described in ref. 22, except that the samples in the current study were separated using 7.5–15% gradient SDS/PAGE. We used affinity-purified anti-Acp antibodies [CG1656, CG1652, and CG17575 (22)], partially purified anti-serum [for SP (17); gift of Eric Kubli], or anti-serum (for CG9997; see SI Text). We have antibodies specific to CG1652 or CG1656; therefore, we represent their results independently, even though both are knocked down simultaneously by RNAi.

Sperm Counts.

Sperm counts were done as in ref. 43. Slides were coded before counting and decoded afterward to avoid bias. Each slide was counted twice; repeatability was >91%. Five replicates were counted per treatment, and data were analyzed using a two-tailed Student's t test.

Immunofluorescence.

Immunofluorescence to detect sperm-bound Acps was carried out following the protocols of refs. 14 and 44, with minor modifications. Please see SI Text for details.