A cis-regulatory Sequence Within the yellow Locus of Drosophila melanogaster Required for Normal Male Mating Success

Abstract

Drosophila melanogaster males perform a courtship ritual consisting of a series of dependent fixed-action patterns. The yellow (y) gene is required for normal male courtship behavior and subsequent mating success. To better characterize the requirement for y in the manifestation of innate male sexual behavior, we measured the male mating success (MMS) of 12 hypomorphic y mutants and matched-outbred-background controls using a y⁺ rescue element on a freely segregating minichromosome. We found that 4 hypomorphs significantly reduced MMS to varying degrees. Reduced MMS was largely independent of adult pigmentation patterns. These mutations defined a 300-bp regulatory region upstream of the transcription start, the mating-success regulatory sequence (MRS), whose function is required for normal MMS. Visualization of gene action via GFP and a Yellow antibody suggests that the MRS directs y transcription in a small number of cells in the third instar CNS, the developmental stage previously implicated in the role of y with regard to male courtship behavior. The presence of Yellow protein in these cells positively correlates with MMS in a subset of mutants. The MRS contains a regulatory sequence controlling larval pigmentation and a 35-bp sequence that is highly conserved within the genus Drosophila and is predicted to bind known transcription factors.

REPRODUCTION usually involves heterosexual courtship behavior that is central to the divergence and diversity of animal species and is of obvious adaptive significance. In many species the basic program of courtship behavior is innate. These inborn, instinctual behaviors are likely to be a result of gene action during development that establishes the potential for behavior and motivates the animal to perform the behavior, given the appropriate external stimulation (Baker et al. 2001). It can be hypothesized that there are genes required for building into the central nervous system (CNS) the ability to process information specific to a behavior and the specific neural output pathway for signaling the performance of that behavior (Baker et al. 2001).

Individual fruit flies of Drosophila melanogaster routinely perform innate behaviors. One such behavior that is well-characterized is the courtship ritual performed by males for females prior to heterosexual copulation. The male ritual is in essence a series of dependent fixed-action patterns: tapping, following, orienting, horizontal wing extending, wing vibrating (“singing”), genital licking, and attempted copulation (Bastock and Manning 1955; Bastock 1967; Hall et al. 1982; Hall 1994a; Yamamoto et al. 1997; Greenspan and Ferveur 2000). Typically, these behaviors must be performed in the correct sequence with some repetition over a modest period of time (2–10 min) to significantly stimulate a female to be receptive to copulation. Such a stereotypic courtship sequence is common to many animals (e.g., Darwin 1874; Bastock 1956; Morris 1970; Walters 1988; Bradbury and Vehrencamp 1998; Carew 2000; Judson 2002).

Many genes whose functions are required for normal D. melanogaster reproductive behavior have been described (Sturtevant 1915; Hall et al. 1982; Hall 1994a; Greenspan 1997; Yamamoto et al. 1997, 1998; Goodwin 1999; Yamamoto and Nakano 1998, 1999; Gaines et al. 2000; Orgad et al. 2000; Romanova et al. 2000; Greenspan and Ferveur 2000; Sokolowski 2001; Boll and Noll 2002; Dauwalder et al. 2002; Subocheva et al. 2003; Grosjean et al. 2004). However, there is a paucity of information about how these genes function in the development of innate courtship behavior. At present, studies of the fruitless (fru) gene provide most of our information about how genes specify D. melanogaster male courtship behavior. Wild-type fru function is required for most aspects of the male courtship ritual, and therefore of male mating success (MMS) (Ryner et al. 1996; Villella et al. 1997). The fru gene encodes a predicted BTB/POZ family zinc-finger transcription factor, which has been shown to be a member of the D. melanogaster somatic sex determination hierarchy (Ito et al. 1996; Ryner et al. 1996; Heinrichs et al. 1998). The use of an antibody specific to male-limited FRU proteins (FRU^M) (Lee et al. 2000; Lee and Hall 2001) has demonstrated that the spatial and temporal distribution of FRU^M proteins in the CNS is consistent with the timing of the critical period for programming a male fate (Belote and Baker 1987; Arthur et al. 1998) and the regions of the CNS whose maleness is necessary for normal development of male courtship behaviors (Hall 1979; von Schlicher and Hall 1979). Most recently, it has been shown that sex-specific splicing of fru specifies the male courtship behavior fate (Demir and Dickson 2005; Manoli et al. 2005; Stockinger et al. 2005). The available evidence suggests that male-specific FRU transcription factors form the apex of a developmental genetic hierarchy responsible for many aspects of the male courtship ritual and MMS in D. melanogaster (reviewed by Baker et al. 2001; Drapeau 2001a; Greenspan and Ferveur 2000; Sokolowski 2001; Hall 2002).

The direct and indirect targets of FRU^M are largely unknown. We recently demonstrated that the yellow (y) gene is in a pathway downstream of fru in a small number of cells in the third instar CNS (Drapeau et al. 2003). Yellow protein accumulation in the CNS is disrupted in at least two fru mutants, and ectopic FRU^M is sufficient to increase levels of Yellow protein throughout the CNS, even in females (Drapeau et al. 2003). On the basis of this evidence we predict that y is either a direct target of FRU^M or a downstream gene in the pathway.

The y gene is an excellent candidate to study further in relation to the development of male sexual behavior. It has long been known that y is required for normal amounts of wing extension (WE) behavior during male courtship and consequent MMS (e.g., Sturtevant 1915; Bastock 1956; Burnet et al. 1973). Male-limited WE is important because it is performed in D. melanogaster for the sole purpose of singing the courtship song, which stimulates females to be receptive to the male's mating advances (reviewed in Hall 1994a; Greenspan and Ferveur 2000). Males carrying null alleles of y have a reduction of ∼50% normal WE levels, which is likely to contribute to the well-characterized decrease in y mutant MMS (Bastock 1956; Burnet et al. 1973; Drapeau et al. 2003). We found that y⁺ expression at the third instar developmental stage was sufficient to rescue male WE and competitive MMS to normal levels (Drapeau et al. 2003), suggesting that y is crucial for laying the foundations of the elaborate courtship repertoire.

Understanding the relationship between a gene and a behavior requires analysis of behavioral phenotypes of numerous mutants, followed by characterization of the wild-type gene and its product(s) in the manifestation of behavior. Some of the most powerful evidence that the function of a specific gene is required for a behavior is the identification of different mutant alleles that have varying effects on the behavior (e.g., period, Konopka and Benzer 1971; fruitless, Villella et al. 1997; foraging, de Belle et al. 1989). In this report, we characterize 3 null and 12 hypomorphic alleles of yellow and find a spectrum of allelic effects on MMS. Knowing the molecular basis of the hypomorphic lesions allowed us to define a 300-bp regulatory region of y whose function is necessary for normal MMS. We name this region the mating-success regulatory sequence (MRS). On the basis of additional genetic and biochemical experiments, we conclude that the MRS is important for the regulation of y expression in the developing CNS. A 35-bp region of the MRS is highly conserved within the genus Drosophila, and this region is predicted to bind at least one transcription factor involved in somatic sexual differentiation. These data suggest that y plays a critical role in the development and evolution of a complex adult behavior with relevance to organismal fitness.

MATERIALS AND METHODS

Fly culture conditions:

D. melanogaster flies were raised in half-pint glass milk bottles on standard banana/corn syrup/yeast/agar medium at room temperature in constant light. Bottles for experiments were always cleared of parents before 1 week after egg laying, and usually within 3–4 days. Parents were sometimes fed a baker's yeast/distilled water paste while laying eggs, which became progeny for experiments. We note that Dow (1977c) has demonstrated that neither light regime (24:0 L:D constant light vs. a 12:12 L:D regime) nor temperature (15° vs. 20° vs. 25°) has a strong influence on the relatively poor mating success of y¹ vs. wild-type males, assayed with wild-type females.

Mutant yellow strains:

Descriptions of the y mutants used in these experiments, including their full names, abbreviations, origin, molecular genetic phenotypes, and relevant references and sources are summarized in Table 1. The mutants include spontaneous mutants with lesions in the endogenous y locus and mutants constructed in vitro and then inserted into a y null background using standard germline transformation protocols (e.g., Rubin and Spradling 1982). All the mutant stocks are highly inbred, but experiments were carried out in completely outbred genetic backgrounds; see below.

TABLE 1.

Descriptions of the alleles used in this study

yellow allele	Full allele name	Molecular genetic basis	Endogenous or transgene	Pigmentation allele class	Source	Representative references
y1	yellow 1	Point mutation (A-to-C transversion) in translation initiation codon	Endogenous	Null	Bloomingtona	Geyer et al. (1990)
Df(1)y-ac22	Deficiency(1) yellow achaete 22	Deletion covering entire y locus	Endogenous	Null	Bloomington	Biessmann and Mason (1988)
y59b	yellow 59b	Deletion of 1.2 kb from −700 to +519 bp and insertion of 3.2 kb of gypsy element at deletion site	Endogenous	Null	Bloomington	Geyer et al. (1990)
y−893	yellow -893	Insertion of insulating vector at −893 bp	Transgene	Hypomorph	Victor Corces	Geyer and Corces (1992)
y2	yellow 2	Insertion of full-length gypsy element at −700 bp	Endogenous	Hypomorph	Bloomington	Campuzano et al. (1985); Chia et al. (1986); Parkhurst and Corces (1986)
y3P	Inversion(1) yellow 3 of Patterson	Inversion with left inversion breakpoint at ∼−700 bp	Endogenous	Hypomorph	Bloomington	Campuzanoet al. (1985); Biessman (1985)
y−300gin	yellow -300 gypsy in	Gene replacement: insertion of insulating vector at −300 bp	Transgene	Hypomorph	Pamela Geyer	Parnell et al. (2003)
ybl	yellow bristle	Three small aberrations 5′ of −700 bp	Endogenous	Hypomorph	Bloomington	Biessman (1985)
y−91	yellow -91	Deletion from −2873 to −91 bp	Transgene	Hypomorph	Pamela Geyer	Geyer and Corces (1987)
y3d	yellow 3 dark	Insertion of ∼7 kb between −300 and 0 + 1 bp	Endogenous	Hypomorph	Bloomington	Chia et al. (1986)
y2S	yellow 2 from Swedish	Insertion of ∼4.5 kb between −300 and +1 bp	Endogenous	Hypomorph	Bloomington	Chia et al. (1986); Parkhurst and Corces (1986)
ya77	yellow a77	Insertion of P element in reverse orientation at −4 bp	Endogenous	Hypomorph	Pamela Geyer	Morris et al. (1999)
yintronless	yellow intronless	Deletion of entire y intron	Transgene	Hypomorph	Pavel Georgiev	Geyer and Corces (1987)
y+660R	yellow +660R	Insertion of insulating vector at +600 bp	Transgene	Hypomorph	Victor Corces	Geyer and Corces (1992)
y+1310	yellow +1310	Insertion of insulating vector at +1310 bp	Transgene	Hypomorph	Victor Corces	Geyer and Corces (1992)

^aBloomington (Indiana) Drosophila Stock Center.

The hypomorphic y mutants that we measured for MMS were chosen because each in some manner eliminated the sequence and/or function of regulatory regions, or potential regulatory regions, of the y locus. For a given mutant, this was accomplished in one of four ways: insertion of foreign sequences, an inversion breakpoint within the y locus, deletion of y sequence, and “gypsy insulation” of regulatory elements. The last method requires a brief explanation. Transposable elements of the gypsy class in D. melanogaster (Lindsley and Zimm 1992; FlyBase 2003) contain binding sequences for the Suppressor-of-Hairy-wing [Su(Hw)] protein. Bound Su(Hw), in combination with other proteins, blocks the interactions of distal enhancers with target promoters and constitutes the gypsy insulator. Except for the spontaneous y² mutant, which contains a full-length gypsy element insertion (Geyer et al. 1986; Parkhurst and Corces 1986), and y^−300gin (see below), all insulator mutants used were constructed by Geyer and Corces (1992). These authors constructed P-element vectors containing 12 Su(Hw)-binding sites inserted at different sites within a y transgene. These Su(Hw)-binding sites are sufficient to function as an insulator. Transgenic stocks were produced by germline transformation of a parental y deficiency line. The y^−300gin allele contains the gypsy insulator positioned 300 bp upstream from the transcription start and was generated at the endogenous y locus using targeted gene replacement (Parnell et al. 2003).

Other Drosophila strains:

To further examine the function of y regulatory sequences, we searched the Gal4 Enhancer Trap Database (GETDB, http://flymap.lab.nig.ac.jp/∼dclust/getdb.html) for a strain carrying a P{GawB} element (Brand and Perrimon 1993) inserted within X chromosomal region 1A5, where y is located. GETDB strain 6014 drives Gal4 expression in tissues that have y-dependent pigmentation (e.g., larval mouth parts, adult dorsal abdominal stripe). This mutant, which we name y⁶⁰¹⁴, contains a P{GawB} element inserted at +23 bp relative to the start of transcription. This ∼11.3-kb P-element insertion causes a severe reduction in larval and adult pigmentation and the absence of anti-Yellow staining in the third instar CNS, suggesting that y⁶⁰¹⁴ is a loss-of-function allele. We used the GAL4 feature within the element to “read out” the surrounding regulatory sequence that may be relevant to the development of behavior. In this experiment, y⁶⁰¹⁴ females were crossed with w; UAS-nu.GFP males, which carry a gene encoding nuclear-localized green fluorescent protein that is responsive to a Gal4 element driving in trans. We observed GFP staining in the third instar larval CNS of y⁶⁰¹⁴; UAS-nu.GFP/+ progeny.

Behavior genetic experimental design:

Because of the sensitivity of behavioral phenotypes to environmental influence, the genetic background of the mutant and of the matched-background control males tested in our mating assays was well controlled. This was a consequence of our experimental design, which incorporated a one-generation crossing scheme. For assays of male sexual behavior, we generated outbred and nearly isogenic experimental and control males using a strain that contains a freely segregating X chromosome-originating duplication, Dp(1;f)8-23 [Dp(1;f)1187, P{ry^+t7.2 = PZ}0801 P{ry^+t7.2 = PZ}8-23 y⁺; In(1)sc⁸, Df(1)sc⁸ (FlyBase 2003)]. Dp(1;f)8-23 is a segregating hemizygous 1.3-Mb “minichromosome” (one of the smallest known to segregate in eukaryotes) that contains ∼1000 kb of X chromosome centromeric heterochromatin and 300 kb of X chromosome distal DNA (e.g., Tower et al. 1993). This distal euchromatic region includes the endogenous y locus and its naturally occurring surrounding regulatory DNA, the achaete (ac) locus, and the scute (sc) locus up to the breakpoint named sc⁸ (Lindsley and Zimm 1992; FlyBase 2003). Distal to y are 10 annotated genes, including 3 genes that have been studied previously, Or1a, cin, and ewg (information from the FlyBase Genome Browser at http://www.fruitfly.org/cgi-bin/annot/gbrowse, analyzing release 3 of the D. melanogaster genome). Finally, Dp(1;f)8-23 has two rosy⁺ (ry⁺) P-element insertions (Figure 1A; Lindsley and Zimm 1992; Tower et al. 1993; FlyBase 2003; Kumar Hari, personal communication).

Figure 1.

(A) Using the yellow⁺ minichromosome Dp(1;f)8-23 to rescue yellow mutants. This minichromosome was created by a duplication of part of the tip of the X chromosome where the y locus resides. Minichromosome 8-23 and related minichromosomes are the smallest chromosomes known to be transmitted in eukaryotes. 8-23 carries the entire endogenous y locus with all of its normal flanking DNA, which includes previously characterized regulatory regions that act as enhancers. Approximately 12 wild-type genes are carried on the minichromosome, as well as two rosy⁺ P-element insertions. The P elements and the yellow and achaete loci are shown. The striped rectangle is ∼1000 kb of centromeric heterochromatin. (B) The crossing scheme used to generate experimental and control males. This one-generation crossing scheme takes advantage of the y⁺ locus located in minichromosome 8-23, and it was used for all of the experiments reported in this article. To generate completely outbred and isogenomic males, we crossed females (with one genetic background, in pink) with the y mutant allele of interest (y⁻) to males (with another genetic background, in blue) that were y¹; ry⁵⁰⁶; 8-23, ry⁺ y⁺. (The y⁺ minichromosome is shown by a black/yellow striped line.). All sons receive the same X chromosome carrying a y mutant from their mother, and ∼75% of the males receive the y⁺ minichromosome from their father (see materials and methods). In this manner, y⁻ (“experimental”) and y⁻ ; y⁺ (“control”) males are generated for analysis. Experimental and control males have the same set of parents and develop in the same set of bottles, and the behavior of both groups is measured simultaneously in the Copulatron.

To generate mutant and control flies, y⁻ (representing any y mutant in Table 1), virgin females were crossed to y¹ sc w^a; ry⁵⁰⁶; Dp(1;f)8-23, y⁺ males. Flies segregating from this cross are y⁻ ; ry⁵⁰⁶/+ and y⁻; ry⁵⁰⁶/+; Dp(1;f)8-23, y⁺ (Figure 1B). The flies lacking the minichromosome have the y⁻ mutant phenotype, and the flies carrying the minichromosome constitute the y⁺ control phenotype in an otherwise identical, outbred background. We will occasionally abbreviate these experimental and control genotypes as y⁻ and y⁺, respectively. A minority of mutant y alleles studied (y⁻⁸⁹³, y^+660R, y⁺¹³¹⁰, and y^intronless) carried transgenes that were transformed into a Df(1)y background and were not located at the endogenous y locus. These lines were homozygous for the dominantly acting inserts, and so progeny were heterozygous for the inserts and their genotypes were formally Df(1)y/Y ; P[y⁻]/+; ry⁵⁰⁶/+ (i.e., y⁻) and Df(1)y/Y; P[y⁻]/+; ry⁵⁰⁶/+; Dp(1;f)8-23, y⁺ (i.e., y⁺).

Germline segregation of the y⁺ minichromosome in our crossing scheme was investigated. We found that the y⁺, Dp(1;f)8-23 minichromosome was transmitted to offspring ∼75% of the time, with no significant sex bias. In crosses of y⁺, Dp(1;f)8-23 males to y¹ females, counts of 922 male and 955 female offspring showed that there were 235 (25.5%) y¹ males and 687 (74.5%) y⁺ males and 243 (25.4%) y¹ females and 712 (74.6%) y⁺ females.

With this protocol, we achieved excellent control over what can collectively be called “background effects,” which have the potential to be misleading in behavioral genetics studies (e.g., de Belle and Heisenberg 1996; Crabbe et al. 1999). Use of the minichromosome crossing scheme allowed us to examine the behavior of outbred “experimental” and “control” male progeny, which were full sibs, developed under the same environmental and developmental conditions, with identical genetic backgrounds except for the presence or absence of the y⁺ minichromosome. Outbred flies with well-controlled background genetic and environmental effects provided excellent material for testing hypotheses associated with the relationship between specific y alleles and male mating success.

Noncompetitive mating behavior observations:

All noncompetitive mating experiments were carried out in a set of behavior observation chambers collectively called the “Copulatron,” the construction of which is described in detail elsewhere (Drapeau and Long 2000). Briefly, the Copulatron is a set of 49 1-inch-diameter Plexiglas circular mating chambers arranged in a 7 × 7 configuration. Each chamber is ∼2 cm deep and can be divided into two 1-cm-deep sections. The floor of the chamber is white porous polyethylene, and the ceiling is glass. Fly media was pipetted into side chambers connected to each of the mating chambers (see Drapeau and Long 2000) so that flies would not starve while recovering from anesthesia and general handling for ∼24 hr.

Male noncompetitive mating success was assayed in the Copulatron by pairing single virgin males with single virgin females. We always assayed experimental and control males simultaneously in the same Copulatron unit. All flies were collected as 6- to 8-hr-old virgins and stored in groups of four to nine in plastic 8-dram food vials for ∼3–4 days. Flies were then briefly (∼30 sec) anesthetized with CO₂ and mouth pipetted, or gently picked up with a paintbrush and placed, into the Copulatron's mating chambers. A single male and female were put in each chamber. Fresh standard fly media was always present in the chambers. Males and females were separated by a lightly lubricated (Pam cooking oil wiped on with paper towels) piece of overdeveloped (i.e., darkened) film, which was removed at the start of each assay, simultaneously introducing males and females in all 49 chambers. Flies were allowed to recover from CO₂ anesthesia for ∼24 hr before the start of each experiment. The flies were exposed to constant light at room temperature.

Because it is inherently difficult to understand the behavior of mutant flies with very poor courtship behavior by using a measurement such as time to copulation, male noncompetitive mating success was defined as follows. On a given day, during which experimental and control males were without exception assayed simultaneously, the rank order of copulations was determined (i.e., the first male to copulate was given the rank 1, the second was assigned rank 2, etc.). Occasionally, two males who effectively began copulation simultaneously were given a rank that was the average of the next two ranks. Within a given assay on a given day, males who mated before the median rank were defined as successful (“S”). Males who mated at or after the median rank were failures (“F”). Numbers of successes and failures for a genotype were denoted as #S and #F. This ranking scheme circumvents the problem of losing data from y males who did not copulate within a reasonable amount of time, which we defined as roughly 1 hr. Normal, healthy, virgin males and females of this species typically achieve copulation within 2–5 min of introduction to each other. We required copulations to last for at least 2 min to be recorded as valid; D. melanogaster copulation duration is typically 10–20 min. (Occasionally, a male will attempt to copulate and will appear to succeed, but the female will not allow the male to fully engage. In this situation the male will climb off the female and nearly always resume courtship of her. This series of events typically takes <2 min.)

We performed independent measurements (trials) of male mating success for each y mutant. Each trial took place on a separate day. To determine whether there was significant variation in mating success estimates across trials within mutants, we carried out chi-square tests on y mutant male #S and #F in an n × 2 format, where n is the number of trials for that mutant. For each of the 15 mutants, we found P > 0.01, with 14 of 15 P-values ≫ 0.05 [The Df(1)y-ac²² flies had P = 0.02, but in all trials such males had extremely reduced mating success; see results.] Because data were quantitatively consistent across trials for all mutants, mating success data for each y mutant and its matched-background control were combined across all trials, and statistical tests of homogeneity were performed on the overall data to test whether mutants had a different distribution of mating success and failure with wild-type females than did controls. For each mutant, three tests were performed on #S and #F for experimental and control males, with data organized into 2 × 2 contingency tables: a Fisher's exact test, a Pearson's chi-square test, and a Mantel-Heanszel chi-square test.

Competitive mating behavior observations:

All competitive mating experiments were performed in 8-dram plastic vials with cotton plugs. The vials contained standard banana/corn syrup/yeast/agar fly media on the bottom. Flies were isolated as 6- to 8-hr-old virgins, and the two sexes were separated and stored in groups of four to seven in food vials at room temperature. At 3–4 days of adult age, control and experimental males were placed in food vials, one of each genotype, using CO₂ anesthesia. We found no evidence for environmental “bottle” effects in pilot experiments in which competing males were placed in a vial together only if they were collected from the same bottle in which they developed (data not shown). Wild-type females were placed singly in food vials. At 4–5 days of adult age, and 24 hr after CO₂ anesthesia (to allow adequate recovery) males were transferred into female vials without anesthesia by “tapping.” This is a simple, commonly used assay to ascertain relative mating success of two genotypes (e.g., Sturtevant 1915; Threlkeld et al. 1974; Markow 1981; Markow and Manning 1982). For each individual female, the male genotype that achieved copulation first was recorded. Male genotypes were differentiated by pigmentation of the body cuticle or eyes. The fraction of trials in which a given genotype achieves the first copulation can be used as a measure of female preference for one genotype relative to the other. Competitive mating experiment data were analyzed using binomial tests to determine whether the distribution of “first copulations” among the two male genotypes deviated from a 50:50 null hypothesis.

Molecular genetic analysis of y insertion mutations:

Three y mutants found to disrupt male mating success, y^3d, y^a77, and y^2S, were partially characterized at the molecular genetic level before this study (Table 1). These three alleles were known to have inserts of unknown origin within the ∼700 bp 5′ of the endogenous y start of transcription: y^3d was defined by an ∼7.0-kb insert, y^a77 was characterized by an insertion of unknown size, and the y^2S mutant was defined by an ∼4.5-kb insert (reviewed in Morris et al. 1999). To confirm the presence of the inserts in this general region, we performed PCR with forward primer 5′-AATATTTGACCCTCAGTGAATTGTG-3′ and reverse primer 5′-TATATGTTCAGCTATAGGGGGTTCTT-3′. In a wild-type fly the resulting amplicon is 888 bp, covering the entire ∼700-bp region of interest, plus flanking sequence on each side. We carried out 10-μl PCR reactions with the following protocol: 4 min at 94°, 35 × [45 sec at 94°, 45 sec at 56°, 4 min at 72°], 1 min 72°, followed by a 4° hold. Using these primers we were able to confirm the presence of the inserted DNA in the y^3d, y^a77, and y^2S mutants. Then we used forward primer 5′-TTCTGGATTTTTGTCTGCATGT-3′ and reverse primers 5′-ATGATTTTTGGCCTTCATCG-3′ and 5′-GCCTTCGGCTGTGTGATATT-3′ to subdivide the −700- to 0-bp region into two sections from ∼−700 to −300 bp and from −300 to +1 bp to further map the locations of the three insertions.

We partially sequenced the male mating success regulatory region of y^a77 to further characterize its insert. The following primers were used both for PCR amplification of fragments and for sequencing: forward primer 5′-GCTCAAAATCACCTGCCAAT-3′ and reverse primer 5′-AGGTGATCAGGGTCACAAGG-3′. The resulting sequence was deposited in GenBank with accession no. DQ099804.

Larval dissection and immunocytochemistry:

Third instar larvae were dissected in 1× phosphate-buffered saline (PBS) using fine forceps, incubated for 15 min at room temperature (RT) in 1× collagenase (Sigma, St. Louis) in PBS, incubated 30 min at RT in 4% paraformaldehyde solution, washed two times for 10 min at RT in PBS/1% Triton, blocked in PBS/0.5% Triton/10% horse serum for 45 min at RT, incubated overnight at 4° with primary antibody, washed three times for 10 min at RT in PBS/0.5% Triton, incubated with secondary antibody for 60 min at RT, washed three times as above, and placed in 50% glycerol solution. The CNS of each larva was then dissected using needles on a glass slide. Polyclonal rabbit anti-Yellow antibody has been previously described (Radovic et al. 2002; Wittkopp et al. 2002a; Drapeau et al. 2003) and was used at a dilution of 1:150. For secondary antibody, a dilution of 1:1000 anti-rabbit 594 (red, Molecular Probes, Eugene, OR) was used. Fluorescent images were taken using a Nikon Microphot-SA fluorescent microscope with a Nikon super-high pressure mercury lamp. Confocal images were taken using a Leica TCS SP2 confocal microscope. Images were processed using Adobe Photoshop on a Macintosh.

Sequencing natural y alleles of D. melanogaster:

To identify naturally occurring genetic variation that might contribute to the evolution of male sexual behavior, we sequenced the MRS (see results) and its flanking sequence in 85 strains of D. melanogaster. These strains were: Oregon-R and Canton-S (both from the Bloomington Drosophila Stock Center), an iso-female line derived from the outbred Ives strain (from Michael Rose, University of California, Irvine, CA), 71 iso-X chromosome strains derived from a recent collection in Napa Valley, California (A. D. Long, unpublished data), 7 iso-X chromosome strains derived from a collection from Zimbabwe (from Chuck Langley, University of California, Davis, CA), and four iso-X chromosome strains derived from a collection from North Carolina (also from Chuck Langley). We used forward primer 5′-GCTCAAAATCACCTGCCAAT-3′ and reverse primer 5′-AGGTGATCAGGGTCACAAGG-3′ to sequence part of the y locus from these strains. These sequences were deposited in GenBank with accession nos. DQ099805–DQ099807 and DQ099812–DQ099893.

Evolutionary analysis of y alleles from diverse Drosophila species:

To study patterns of change and constraint within the MRS over evolutionary time, we sequenced the appropriate y region in a number of Drosophila species. As with the D. melanogaster strains, we used forward primer 5′-GCTCAAAATCACCTGCCAAT-3′ and reverse primer 5′-AGGTGATCAGGGTCACAAGG-3′ to sequence strains of each of its three sibling species, D. simulans, D. sechellia, and D. mauritiana. The strains that we sequenced were: D. simulans sim6 and s132 (from Dave Begun, University of California, Davis, CA), D. sechellia S-9 (from the Arizona Drosophila Species Center, Tucson, AZ), and D. mauritiana iso-197 (Arizona Drosophila Species Center). These sequences were deposited in GenBank with accession nos. DQ099808–DQ099811.

In addition, we used an evolutionarily diverse set of previously sequenced y alleles in our analysis. Species, strain (if available), NCBI sequence identification numbers, and relevant references are as follows: D. simulans, Sim-5 G20, AB026336, Takano-Shimizu (1999); D. orena, AB026342, Takano-Shimizu (1999); D. teissieri, AB026339, Takano-Shimizu (1999); D. erecta, AB017574, Takano-Shimizu (1999); D. yakuba, AB017573, Takano-Shimizu (1999); D. subobscura, from Figure 2 of Munté et al. (1997); D. pseudoobscura from the Baylor School of Medicine D. pseudoobscura Genome Project; D. virilis, AY128944, Wittkopp et al. (2002b); and D. novamexicana, 15010-1031.0, AY165561, Wittkopp et al. (2003b).

Figure 2.

(A) Noncompetitive male mating success of three y null mutants, y¹, Df(1)y-ac²² [here abbreviated Df(1)y], and y^59b. Yellow bars represent behavioral data collected from experimental (mutant) males, and black bars represent data from control males (carrying the minichromosome). Bars represent means + SEM, calculated across behavioral trials. Data were analyzed using Fisher's exact tests, and the P-values from this analysis are shown. Sample sizes and additional statistical analyses are shown in Table 2. (B) Competitive male mating success of the same y null mutants. Raw data converted to percentages of total competitions are shown. Here, mating success is measured as the percentage of time that a genotype achieves first copulation with a female when in direct mate competition with another genotype. Data were analyzed with binomial tests to determine for each mutant-and-control combination whether a significant difference from a 50:50 distribution among males existed. In all cases, we found a highly significant difference (see results).

Bioinformatic predictions of transcription factor binding sites:

Using the 300-bp MRS (see results) as our query, we searched the TRANSFAC database of transcription factor binding sites (Heinemeyer et al. 1998) using the TFSEARCH tool (http://www.cbrc.jp/research/db/TFSEARCH.html) (Akiyama 1998). We also utilized the FLY GRID database of genetic and physical interactions (http://biodata.mshri.on.ca/fly_grid/servlet/SearchPage). Finally, empirically determined transcription factor binding sites from the primary literature were used in our analysis.

RESULTS

Null y mutants have severely reduced male mating success:

The D. melanogaster y¹ allele has been previously characterized as an A-to-C transversion in the first translated codon of the yellow gene, which results in an altered mRNA product (Geyer et al. 1990). Since Sturtevant's pioneering genetic study of Drosophila sexual behavior (Sturtevant 1915) researchers have recurrently shown that y¹ males have severely lowered MMS (Spett 1931; Diederich 1941; Merrell 1949; Mayr 1950; Meyers 1953; Bastock 1956; Barker 1962; Schroeck 1971; Burnet et al. 1973; Threlkeld et al. 1974; Dow 1975, 1977a, b, c, d; Tracey and Espinet 1976, 1977; Wilson et al. 1976; Oakeshott and Hayman 1979; Bradman et al. 1981; Mizuguchi and de Almeida 1983; Heisler 1984; Hamerlynck 1994; Drapeau et al. 2003). The y¹ allele is a null with regard to its pigmentation phenotypes (FlyBase 2003), but it has never been explicitly demonstrated that the canonical “poor male mating success” y mutant, y¹, meets the definition of a null/amorphic allele. A more specific question is whether the y¹ mRNA has a neomorphic function in y¹ males that alters their courtship behavior and MMS. An alternative hypothesis is that the y¹ mRNA is nonfunctional.

Because y is on the X chromosome and males normally have only a single copy, y¹ will be defined as a null if the y¹ MMS phenotype is equivalent to that of a male carrying a single X-linked deletion of the y locus (Muller 1932). Using the y⁺ minichromosome rescue technique (Figure 1), we measured the noncompetitive MMS of y¹ males as well as of males carrying either of two null mutants of y, which have been characterized as deletions: Df(1)y-ac²², a deletion of the entire y locus, and y^59b, which is a deletion of genomic sequence −700 to +519 bp relative to the start of transcription, with an insertion of 3.2 kb of gypsy retrotransposon sequence (Table 1).

In noncompetitive mating assays, males carrying any of these three mutants had significantly worse mating success than their matched-background control males carrying the y⁺ minichromosome, with statistical significance in each case determined by 2 × 2 tests of homogeneity within mutants (Table 2; Figure 2A). We additionally evaluated the behavior of males carrying these three mutants in a distinct competitive MMS paradigm. Consistent with the noncompetitive mating test results from the preceding experiment, we found that all three mutants had severely and significantly reduced mating success in a one-mutant male, one-control male competitive situation in a modest-sized arena, with statistical significance in each case determined by a binomial test for a deviation from a 50:50 distribution of y¹:y⁺ competitive MMS (Figure 2B).

TABLE 2.

Statistical tests of homogeneity of yellow mutant and matched-control noncompetitive male mating success

yellow allele	Total Na	No. of trials	% success in experimental flies	% success control flies	Fisher's exact test P-valuebc	Pearson's χ2 d	Pearson's P-valuec	Mantel- Heanszel chi-squarede	Mantel- Heanszel P-valuec	Overall male-mating success
y1	120	5	32.9	74.0	8 × 10−6	19.75	< 0.0001	19.58	< 0.0001	−
Df(1)y-ac22	83	4	18.2	76.9	6 × 10−8	28.74	< 0.0001	28.39	< 0.0001	−
y59b	82	3	13.2	79.5	9 × 10−9	35.97	< 0.0001	35.53	< 0.0001	−
y−893	123	4	45.3	58.6	0.10	2.14	0.14	2.12	0.15	+
y2	139	5	51.3	47.6	0.73	0.19	0.66	0.19	0.67	+
y3P	70	2	46.7	50.0	0.49	0.08	0.78	0.08	0.78	+
y−300gin	126	4	46.8	48.4	0.50	0.04	0.85	0.04	0.85	+
ybl	193	6	51.5	49.0	0.69	0.13	0.72	0.13	0.72	+
y−91	64	2	17.6	90.0	7 × 10−6	20.13	< 0.0001	19.81	< 0.0001	−
y3d	174	6	23.5	73.0	4 × 10−11	42.64	< 0.0001	42.32	< 0.0001	−
y2S	335	9	44.9	54.8	0.045	3.25	0.07	3.24	0.07	−
ya77	121	4	29.4	64.3	1 × 10−4	14.35	< 0.0001	14.33	0.0001	−
yintronless	166	5	51.4	52.2	0.52	0.01	0.92	0.01	0.92	+
y+660R	88	2	45.8	55.0	0.26	0.73	0.39	0.73	0.39	+
y+1310	109	5	55.2	47.1	0.85	0.72	0.39	0.71	0.40	+

^aTotal N is the sum of the number of experimental and matched-control flies analyzed across all trials.

^bOne-sided Fisher's exact tests were performed because we expected a priori that the mating success of mutant males would either not change or be reduced. The P-value is the probability of the exact contingency table or a more extreme one in the same direction.

^cThe P-values that are <0.05 are underlined, and P-values <0.10 but >0.05 are in italics.

^dAll chi-square tests were performed with 2 × 2 contingency tables and 1 d.f. with a critical chi-square value of 3.841 for a 0.05 significance level.

^eThe Mantel-Heanszel chi-square test is frequently used to analyze data tables in which one or more cells have low values. We utilized this as an additional test because low mating success of some y mutants occasionally resulted in low values for “success,” and/or very high mating success of some control males resulted in low values for “failure.”

Statistical significance was extremely high (P < 0.0001) in each of the six experiments presented above, suggesting that y¹ acts as a null allele with respect to MMS. To more rigorously determine whether the effects of y¹ on MMS are equivalent to those caused by deletions of the gene [Df(1)y-ac²² and y^59b], we carried out additional statistical analyses on the two overall data sets produced from the noncompetitive and competitive mating experiments above. First, we performed two-way ANOVA analysis of arcsine-square root-transformed noncompetitive MMS data with factors “allele” and “minichromosome” (Sokal and Rohlf 1995). Presence or absence of the minichromosome had a highly significant effect on male mating success, as expected since the y⁺ minichromosome rescued poor mutant male mating success (F_1,18 = 52.7, P < 0.0001). Within alleles, discounting the presence/absence of the y⁺ rescue chromosome, there were both successes and failures, and therefore the mutant alleles did not by themselves significantly contribute to the observed variation in overall male mating success (F_2,18 = 0.96, P = 0.40). However, the important part of the ANOVA is the interaction between alleles and the minichromosome, which answers the following question: Did the difference between the presence or absence of the minichromosome fluctuate with varying alleles? In this case, as expected, if all three mutant alleles are null and hence equivalent to each other, the interaction was not significant (F_2,18 = 1.20, P = 0.55). ANOVA analysis of nontransformed data yielded statistically equivalent results (not shown). Second, we performed a 3 × 2 chi-square test on the distribution of y¹ vs. y⁺ first copulations among the three alleles in competitive mating assays and found no evidence for meaningful variation among alleles (χ² = 0.736, d.f. = 2, P = 0.692). Overall, we conclude that y¹ is a null allele with regard to MMS and that lesions in the y locus that reduce MMS in a simple noncompetitive assay are likely to have effects in more ecologically relevant, competitive situations.

Hypomorphic y alleles have a spectrum of effects on male mating success:

The experiments with null alleles presented above provide a foundation for further dissection of the y locus and analysis of modest molecular perturbations on male sexual behavior. We quantified the noncompetitive mating success of males carrying each of 12 previously characterized hypomorphic y alleles (Table 1) to identify alleles that significantly reduced MMS relative to control males carrying the same mutation plus the rescuing y⁺ gene on the minichromosome (Figure 1).

Results from these behavior assays are presented in Table 2 and Figure 3. We found that 8 of 12 hypomorphs had MMS that was statistically equivalent to controls (Table 2). In our behavioral assay, this was manifested as ∼50% mating success for both the mutant and control genotype (Figure 3, gray bars). To determine whether significant variation in MMS exists among this set of eight alleles, we performed two-way ANOVA analysis with factors “allele” and “minichromosome” on arcsine-square root-transformed data, as for the null alleles above. In this case, as expected, we found no significant main effects of alleles or presence of the minichromosome on overall variation in male mating success, and the interaction term was also not significant (all P > 0.05). This statistical outcome is consistent with the following hypotheses: These alleles as a group do not significantly reduce MMS, and the mean mating successes of the different mutant alleles do not significantly differ from each other (they are all ∼50%; Figure 3).

Figure 3.

Noncompetitive mating success of males carrying y hypomorphic regulatory mutations. The y allele name is centered over the appropriate bar depicting the mean ± SEM of that genotype's mating success, calculated across behavioral trials. Mutant male data are represented by yellow (gray) bars if they had significantly lowered (unchanged) mating success, as determined using Fisher's exact tests. Sample sizes and additional statistical analyses are shown in Table 2. Control male data are also shown in Table 2.

In contrast to the eight mutants discussed above, four y hypomorphs, y⁻⁹¹, y^3d, y^a77, and y^2S, caused significant reductions in MMS (Table 2, Figure 3, yellow bars). The degree to which each mutation significantly lowered MMS varied from (mean ± SEM of percentage success) 20.6 ± 2.9 for y⁻⁹¹ to 41.8 ± 4.9 for y^2S. To further investigate this variation in MMS, we performed two-way ANOVA analysis (with factors “mutant allele” and “minichromosome”) of arcsine-square root-transformed MMS data from these four hypomorphs. We tested the hypothesis that although each individual allele lowered MMS, there was significant variation among alleles in their effects on this behavior. As expected, the presence or absence of the minichromosome had a highly significant effect on MMS, because without exception the y⁺ minichromosome rescued poor MMS of the mutants (F_1,34 = 66.4, P < 0.0001). Again, the y mutant alleles by themselves did not significantly contribute to the observed variation in overall MMS (F_3,34 = 0.20, P = 0.89), because within a given allele across the presence/absence of the y⁺ rescue chromosome, there were both successes and failures. Differing from the set of null alleles, we found a highly significant interaction between allele and minichromosome (F_3,34 = 6.7, P = 0.001), as expected if there is significant variation among alleles for the difference in MMS between males with and without the y⁺ rescue chromosome. Identical ANOVA analysis of nontransformed data yielded statistically equivalent results (not shown). We conclude that different molecular perturbations of y can lead to a spectrum of effects on noncompetitive MMS.

Additional statistical analyses were carried out to test whether each of these four alleles should be classified as null or hypomorphic with regard to MMS. We determined whether the number of successes and failures differed between the two deletions Df(1)y-ac²² and y^59b and the four pigmentation hypomorphs: Nonsignificance suggests that the allele acts as a null for MMS, and a significant difference indicates that the allele is a hypomorph. Using Fisher's exact tests, we found statistical support for the following order of allelic effects on reduction in MMS: Df(1)y-ac²² = y^59b = y⁻⁹¹ ≥ y^3d > y^a77 ≥ y^2S. Specifically, we found that Df(1)y-ac²² and y^59b MMS were statistically equivalent and that the MMS of y⁻⁹¹ and y^3d were comparable to each of the nulls (P ≫ 0.05 in all comparisons). Analyses of y^a77 [P < 0.00001 (Df(1)y) and P < 0.10 (y^59b)] and y^2S (P < 0.001 in both comparisons) show that while mutant MMS is significantly lower than that of their respective controls, their relative MMS is greater than that caused by null mutants of the y gene. We conclude that y⁻⁹¹ and y^3d are null with regard to MMS and that y^a77 and y^2S are hypomorphs for MMS. On the basis of the marginally significant differences in mating success of y^2S males in the original analysis (Table 2), this allele has the weakest effect on MMS. Although in a number of individual trials y^2S males had very low MMS, in some trials they were equivalent to control males (not shown). Indeed, detection of an overall statistically significant difference in MMS between y^2S mutants and controls was possible only by assaying hundreds of males (Table 2).

Genetic dissection of y regulatory sequences reveals 300 bp required for male mating success:

The y locus is relatively simple, containing two exons, one intron, and a single transcript and protein product (Geyer et al. 1986; FlyBase 2003). Six general regions of the y locus might contain regulatory sequences affecting MMS: (1) the ∼4-kb noncoding region upstream of the start of transcription, (2) the 5′ untranslated region (UTR), (3) the first exon, (4) the large single intron, (5) the second exon, and (6) the 3′-UTR and sequences downstream of the end of transcription. Because sequences regulating the development of cuticular melanization have previously been identified both in the upstream sequence and in the intron, the regions that we inspected were already known to contain sequences with regulatory function (Beissmann 1985; Geyer and Corces 1987, 1992; Martin et al. 1989). Therefore, these are excellent candidate regions within which to search for sequences regulating the function of y with respect to male-specific behavior.

The behavioral phenotypes of the 12 hypomorphic mutants combined with information about their molecular lesions allowed us to map a 300-bp region of y whose function is required for normal MMS. We named this region the MRS (Figure 4). This conclusion is based on the following observations.

Figure 4.

Summary of behavior genetic data relevant to the mapping of an ∼300-bp regulatory region whose function is necessary for normal male mating success. Numbers labeling the y locus are base pairs relative to the start of transcription (at +1 bp). Locations of pigmentation enhancers (purple ovals) are taken from Geyer and Corces (1987, 1992) and Martin et al. (1989). W, wing; B, body; L, larval mouth parts; Br, bristles; A, aristae; D, larval denticle belts; T, tarsal claws. Boxes indicating conserved noncoding sequence within the MRS are based on Munte et al. (1997) and on analysis in the results and in Figure 7. Most of the remaining noncoding sequence in y is not well conserved (see Munte et al. 1997). Mutations shown in gray have no effect on male mating success, and mutations shown with black-and-yellow stripes significantly lowered male noncompetitive mating success. A square on the outer edge of a mutational effect region indicates that the effect of the insulator mutation extends past the drawing, to the extent of the transgene sequence, or in the case of the y² mutation at the endogenous location, beyond the y locus to intergenic sequence. For clarity, and because the omitted mutations provide redundant mapping information, not all mutations phenotyped in this study are shown. The dashed lines extending from insertions y^3d and y^2S indicate that the precise sites of insertion are unknown, although they have been localized to the −300 to +1 bp region. The sizes of insertions y^3d and y^2S are ∼7 and ∼4.5 kb, respectively. Partial sequencing of insertion y^a77 revealed 357 bp of one end of a typical P element.

First, males carrying three hypomorphic mutants with disruptions in the y intron and the sequence downstream of this intron, y^intronless, y^+660R, and y⁺¹³¹⁰, had normal MMS (Table 2, Figure 3). All three mutants carry transgenes not located at the site of the endogenous y locus. The y^intronless flies contain a y transgene deleted for the ∼2.7-kb intron (Geyer and Corces 1987). Mutant y^+660R insulates regulatory sequence function downstream of +660 bp, and in an identical manner y⁺¹³¹⁰ insulates regulatory sequence function in the intron sequence downstream of +1310 bp. Both lines contain the gypsy insulator within the intron, which begins at nucleotide +409 and ends at nucleotide +3129. Since the y^+660R and y⁺¹³¹⁰ insulators disrupt the function of any regulatory sequence downstream of their respective insertion sites, these data suggest that, within the limits of the y transgene construct, there is no MMS regulatory sequence located in the y coding sequence downstream of the intron (see Geyer and Corces 1992). This includes the second y exon, the 3′-UTR, and any downstream regulatory sequences.

Second, males carrying five hypomorphic mutants with disruptions upstream of the y transcription start, y², y^bl, y^3P, y⁻⁸⁹³, and y^−300gin, had normal MMS (Table 2, Figure 3). Four of these mutants were generated within the endogenous y locus: a full-length insertion of gypsy in y², a mutant characterized by a number of small aberrations named y^bl, an inversion with a single breakpoint within y named y^3P, and a gene-replacement mutant named y^−300gin, which utilizes the gypsy insulator at −300 bp relative to the y transcription start (Parnell et al. 2003). The remaining mutant, y⁻⁸⁹³, carries a genetically engineered transgene of the y locus utilizing the gypsy insulator (Geyer and Corces 1992). The molecular lesions associated with these five mutations are upstream of −300 bp relative to the y start of transcription. Therefore, these data suggest that there is no regulatory region required for normal MMS located in the ∼7 kb of the y locus upstream of −300 bp of the y transcription start site. [We note that while a Fisher's exact test applied to the y⁻⁸⁹³ data resulted in a marginally significant difference of P = 0.10, both chi-square analyses had P-values greater than this (Table 2).] Combined with the nonsignificant results from analysis of the other alleles affecting the same general region of the y locus, for example, y^−300gin, we parsimoniously conclude that the y⁻⁸⁹³ lesion caused no meaningful decrease of MMS. It may be the case that there is a position effect of the transgene in the y⁻⁸⁹³ strain that lowers MMS (see discussion).

Third, males carrying the y⁻⁹¹ mutant, a deletion of bases −2873 to −91 in the noncoding region upstream of the y transcription start within a transgene (Geyer and Corces 1987), had a significant reduction in MMS relative to matched control males (Table 2, Figure 3). This result suggests that the 2782-bp deletion contains a sequence required for normal male mating success. On the basis of this and results for other y mutants already presented, we infer that the 209-bp sequence between −300 and −91 bp of the y locus is required for normal MMS.

Fourth, three insertion mutations upstream of the first y exon reduce MMS. The reduced MMS of y⁻⁹¹ males predicts that insertional mutations in or near the −300- to −91-bp region should reduce MMS. Three mutations of the endogenous y locus—y^3d, y^a77, and y^2S—each were previously shown to be large DNA insertions in the ∼700-bp region just upstream of y transcription start (Morris et al. 1999). These mutants all display a reduction of pigmentation and are classified as hypomorphic because in each case at least one body part with y-dependent pigmentation has normal pigment levels (see Table 3 in Morris et al. 1999). The fact that males carrying each of these y lesions had significantly lower mating success than control males did (Table 2, Figure 3) suggests that these insertions disrupt normal MRS function. They appear to do so to different degrees, however (see below).

TABLE 3.

Adult pigmentation phenotypes caused by yellow mutants are not predictive of male mating success

yellow allele	Wingab	Body	Bristles	Aristae	Tarsal claws	Sex combs	Overall mating successd
y1	−	−	−	−	−	+/−c	−
Df(1)y-ac22	−	−	−	−	−	+/−	−
y59b	−	−	−	−	−	+/−	−
y−893	−	−	+	+	+	+	+
y2	−	−	+	+	+	+	+
y3P	+/−	−	+	+	+	+	+
y−300gin	+	−	+	+	+	+	+
ybl	+	+/−	−	−	+	+	+
y−91	−	−	+	+	+/−	+	−
y3d	+	−	−	−	+/−	+	−
y2S	−	+/−	+/−	+	+	+/−	−
ya77	−	−	−	−	+/−	+/−	−
yintronless	+	+	+/−	−	−	+/−	+
y+660R	+	+	−	−	−	+	+
y+1310	+	+	+	+	−	+	+

^aPigmentation data are a summary of personal observations by M. D. Drapeau and P. K. Geyer, the primary references in Table 1, Lindsley and Zimm (1992), and FlyBase.

^bWing, body, bristles, aristae, tarsal claws, and sex combs are adult pigmented structures.

^c“+/−” indicates that the pigmentation is neither completely wild type (dark) nor mutant (yellow), but rather intermediate.

^dThe male mating success data are from this report.

The reduced MMS caused by y^3d, y^a77, and y^2S predicts that the insertions, before this study known only to be within ∼700 bp upstream of the y transcription start, should be located between −300 bp and the start of transcription at +1 bp. Using PCR analysis the location of each of these insertions was narrowed to the −300- to 0-bp region that contains the MRS (not shown). Partial sequencing indicated that the y^a77 allele had an insertion at −4 bp (see materials and methods). Comparison of this y^a77 insertion DNA to sequences in public databases using BLASTn (Altschul et al. 1990) revealed 94% identity to the 5′-end of a typical P element, including the requisite, conserved, perfect 31-bp terminal inverted repeat region. We conclude that y^a77 is a P-element insertion in reverse orientation to y at −4 bp in the y locus.

Adult male pigmentation is not predictive of mating success:

The most conspicuous trait that y mutant adult males have is reduced pigmentation resulting in a yellow body color (Lindsley and Grell 1968; Lindsley and Zimm 1992). Indeed, we utilized pigment differences between mutant and control male cuticles as a neutral marker to distinguish them for experimentation. However, changes in male cuticular coloration may not be neutral and could in fact reduce mating success, an hypothesis that our data allow us to test. One way in which cuticular pigmentation could influence male mating success is if the female responds to visual cues from the male's body pigmentation during courtship. If she receives less “pigment signal,” she will be less stimulated toward a receptivity threshold, thereby lowering the mating success of males with reduced pigmentation. Wing coloration is of particular interest, because wing colors, patterns, and movements in many insects, including Drosophila, are used as visual courtship signals (e.g., Singh and Chatterjee 1987; True et al. 1999; Kopp and True 2002; Gompel et al. 2005; Robertson and Monteiro 2005).

Our data do not support the hypothesis that variation in adult male cuticle features changes MMS (Table 3). Males carrying the pigment-mosaic (normal pigment in some body parts/tissues, and mutant in other) y², y⁻⁸⁹³, and y^−300gin alleles had greatly reduced pigmentation throughout their entire body and wings, yet their mating success was not meaningfully reduced in comparison to control males (Tables 2 and 3, Figure 3). Furthermore, the pigmentation of other, smaller adult male body parts that females might visually assess during courtship is unrelated to male mating success. For example, mutants y^bl, y^+600R, and y^intronless all have reduced pigmentation in the aristae yet exhibited normal MMS (Tables 2 and 3, Figure 3). These data suggest that females are not using reduced cuticle color as a visual signal of poor male quality and subsequently reducing their receptivity.

There is a second mechanism by which pigmentation of adult male body parts could affect male mating success. Because y null mutant males have reduced melanin in their wings, and most likely reduced sclerotin (a hardening compound found in the cuticle and metabolized from tyrosine in the same biochemical pathway as melanin; Burnet and Connolly 1974; Wright 1987), some physical properties of the wing might be altered such that these flies are less efficient at transmitting auditory (song) or olfactory (pheromone) signals to females while courting. Our data allowed us to test, and reject, the hypothesis that wing cuticle structure variation among y mutants is predictive of male mating success. The normal MMS of y alleles conferring yellow-colored wings (y², y⁻⁸⁹³, and y^−300gin) provides evidence against this hypothesis (Tables 2 and 3, Figure 3). Conversely, y^3d males have normal-colored (gray) wings yet severely reduced mating success (Tables 2 and 3, Figure 3). In total, our data suggest that the adult pigmentation and male sexual behavior phenotypes that y confers are independent of each other.

The MRS overlaps a sequence controlling larval pigmentation:

The ∼300-bp MRS overlaps a regulatory sequence previously characterized as a control region for larval pigmentation (Geyer and Corces 1987; Martin et al. 1989; Figure 4). Not surprisingly, the hypomorphic mutants that disrupt this region (y⁻⁹¹, y^3d, y^a77, and y^2S) reduce both MMS (this article) and larval mouth-part pigmentation (Biessmann 1985; Chia et al. 1986; Geyer and Corces 1987; Harrison et al. 1989; FlyBase 2003; M. D. Drapeau and P. K. Geyer, unpublished observations; see also Martin et al. 1989). Interestingly, as with MMS, these mutants reduce larval mouth-part pigmentation to different degrees, with colors ranging from brown, to golden-brown, to yellow. The larval pigment phenotype and the adult male sexual behavior phenotype may therefore be developmentally linked. However, we reason that the pigmentation state of the larval mouth parts is unlikely to have any direct bearing on female mate choice decisions.

Expression of y in the third instar CNS of wild-type and mutant flies:

Previous work suggested that expression of y, and hence presence of the Yellow protein product, is necessary and sufficient in the third instar CNS for normal male courtship and subsequent mating success (Drapeau et al. 2003; M. D. Drapeau, M. Suster, A. Piccin, and A. D. Long, unpublished data). Using an anti-Yellow antibody applied to wild-type flies, we were able to detect Yellow presence in the cytoplasm of a limited number of cells at this stage (Figure 5). This is generally consistent with our earlier work on the third instar CNS (Drapeau et al. 2003), the detection of cytoplasmic Yellow in cuticle cells (Figure 3 in Wittkopp et al. 2002a), and the fact that Yellow is a secreted protein (Kornezos and Chia 1992; Wittkopp et al. 2002a; Drapeau 2003; Drapeau et al. 2003). We did not observe Yellow in these cells in the loss-of-function y¹ mutant (Figure 5, B and E). Therefore the fluorescence of the anti-Yellow antibody is representative of y expression in this tissue at this developmental stage.

Figure 5.

Presence of Yellow protein in third instar CNS neural cells is associated with adult male mating success. In all images, anti-Yellow antibody fluorescence is shown in red. Shown are fluorescent microscope (A, B, D–I) or confocal microscope (J and K) images of the third instar larval CNS. (A) Canton-S wild-type positive-control CNS, showing anti-Yellow staining in four CNS cells. (B) The y¹ loss-of-function negative-control CNS, showing no anti-Yellow staining. (C) Schematic of the third instar CNS, after Truman et al. (1993, their Figure 7), showing approximate locations of dividing neuroblasts at this developmental stage. The Yellow⁺ cells may be the four neuroblasts in red (see also Drapeau et al. 2003). (D–I) Shown are the CNSs from positive- and negative-control flies, and four hypomorphic y mutants, at higher magnification than that in A and B. (D) Canton-S CNS showing anti-Yellow cells. (E) y¹ CNS showing no anti-Yellow cells. (F) The y⁻⁸⁹³ mutant has moderately low MMS and no anti-Yellow cells in the CNS at this stage. (G) The y^−300gin mutant has normal MMS and normal anti-Yellow cells in the CNS at this stage. (H and I) Both y^3d and y⁻⁹¹ mutants have severely reduced MMS and no anti-Yellow cells in the CNS at this stage. (J and K) Confocal images of anti-Yellow cells from Canton-S positive-control flies, and the y^−300gin hypomorph whose males exhibit normal MMS. Yellow appears to be largely, if not completely, cytoplasmic in these cells.

To test the hypothesis that Yellow presence in these cells is relevant to adult male mating success, we examined Yellow in the third instar CNS of four hypomorphs with varying degrees of MMS (Figures 3 and 5). Mutants y⁻⁹¹ and y^3d have greatly reduced MMS, y^−300gin has normal MMS, and y⁻⁸⁹³ has nonsignificant but moderately low MMS. Yellow presence was associated with MMS levels (Figure 5). We detected a strong cytoplasmic Yellow signal in the y^−300gin mutant, which has normal MMS, as expected (Figure 5, G and K). In contrast, the mutants with severely reduced MMS, y⁻⁹¹ and y^3d, had no Yellow in the third instar CNS, as we predicted (Figure 5, H and I). We found that the y⁻⁸⁹³ mutant, with moderately low MMS compared to control males (P = 0.10, Fisher's exact test; see also Table 2), had no detectable Yellow in the CNS at this developmental stage (Figure 5F), although this may be caused by a transgene position effect (see discussion). In total, the data suggest that the Yellow that we detected in the third instar CNS is most likely related to adult MMS.

The Yellow protein data presented above, in combination with our behavior genetic mapping data, predict that the MRS controls y expression in these third instar CNS cells. To test this, we utilized a novel y mutant, which we named y⁶⁰¹⁴ (see materials and methods). This mutation is caused by a P{Gal4} element inserted ∼23 bp downstream of the endogenous MRS and the y transcription start site (Figure 6A). Combined with a UAS-GFP responder element, we can use y⁶⁰¹⁴ as a reporter of y expression controlled by the MRS and sequences upstream of it. As predicted, we found y⁶⁰¹⁴-controlled GFP expression in the identical location and cell number as that seen when viewing Yellow using immunocytochemistry (Figure 6, B and C). We also saw a small amount of GFP expression in other cells, in no distinct or obvious pattern (Figure 6B; M. D. Drapeau and S. A. Cyran, unpublished results). Unfortunately, because y⁶⁰¹⁴ appears to be a null mutant (see materials and methods), we could not easily colocalize Yellow and y⁶⁰¹⁴-controlled GFP expression in these flies. These data nonetheless strongly suggest that the MRS controls y expression in a small number of third instar CNS cells and that Yellow action in these cells is required for normal adult MMS.

Figure 6.

Evidence that the MRS drives y expression in a small number of cells in the third instar CNS. (A) Diagram of the y locus in the newly described mutant y⁶⁰¹⁴. The y⁶⁰¹⁴ mutation is caused by a P{GawB} insertion at +23 bp relative to the start of y transcription. This P-element insertion contains the Gal4 sequence, allowing sequences upstream of the element, including the MRS, to control expression of a UAS-GFP reporter in F₁ offspring. (B) Fluorescent microscope image of a third instar CNS from the genotype y⁶⁰¹⁴; UAS-nu.GFP/+ with the P{GawB} driving nuclear-localized GFP. We observed expression in four cells in the same location as the Yellow⁺ cells shown in Figure 5 (open arrowhead). Ectopic expression of GFP in a few cells throughout the CNS was also seen (see results). Generally, the strongest GFP expression was seen in the four putative Yellow⁺ cells. (C) Magnification of the four cells next to the white arrow in B.

Natural genetic variation within the D. melanogaster MRS:

To identify naturally occurring genetic variation of potential relevance to the evolution of courtship behavior, we sequenced the MRS of 85 wild-type D. melanogaster strains that have diverse origins and ages since laboratory domestication. The sequenced isolates were the well-characterized laboratory strains Oregon-R and Canton-S (isolated many decades ago in Oregon and Ohio, respectively), Ives (isolated ∼35 years ago in Massachusetts), and strains from Zimbabwe, North Carolina, and Napa Valley, California (isolated within the last decade).

We found no common single nucleotide polymorphism (SNP) within the 209-bp sequence that is strictly required for normal MMS, as defined by the y⁻⁹¹ mutant. There was only a single common genetic variant within the longer 300 bp MRS, an A/C SNP at −36 bp relative to the start of y transcription (C in 22/85 lines = 25.9%). In addition to the single common polymorphism, we identified three rare polymorphisms (<5%) within the MRS. They were an A/G site at −67 [G in two strains (2.4%), both from Napa Valley], a C/T site at −141 [T in three strains (3.5%), all from Zimbabwe], and an A/G site at −225 [G in one strain (1.2%) from Zimbabwe].

Molecular evolution of the MRS within the genus Drosophila:

Males of most Drosophila species perform courtship behavior whose purpose is to stimulate females to be receptive to copulation (Spieth 1952, 1974). While the precise details of the male courtship rituals have diverged among species, similar characteristics such as courtship song have been retained in many cases (see, for example, Table 1 in Spieth 1974). Since there have been numerous reports of males in various Drosophila species carrying null or nearly null mutations at the yellow locus with reduced mating success (Rendel 1944, D. subobscura; Tan 1946, D. pseudoobscura; Frias and Lamborot 1970, D. gaucha; Pruzan-Hotchkiss et al. 1992, D. pseudoobscura; da Silva et al. 2005, D. willistoni; W. J. Etges, unpublished data, D. mojavensis), the general requirement for y in the development of courtship behavior is fairly conserved within the genus. It could be that small changes in the regulation of y have resulted in minor changes in male courtship signals in different species (e.g., compare the sibling species D. melanogaster, D. simulans, D. sechellia, and D. mauritiana, Cobb and Ferveur 1996). Molecular evolution within the 300-bp MRS of the y locus may have consequences for the evolution of courtship behavior within species and reproductive isolation between species.

Noncoding sequence conservation among divergent species can indicate conserved regulatory function (e.g., Moses et al. 2003; Sinha et al. 2004). We determined the patterns of constraint and divergence of the MRS region in Drosophila species with divergent patterns of wing usage during courtship. These species were: D. simulans, D. mauritiana, D. sechellia, D. pseudoobscura, D. subobscura, D. virilis, and D. novamexicana, the most divergent of which has been separated from D. melanogaster for ∼65 million years (MY). Generally, the MRS is poorly conserved across wide species boundaries within Drosophila (e.g., Figure 7B). However, a striking feature of MRS evolution is an extremely well-conserved 35-bp sequence, positioned in the center of the region that we have shown to be required for normal male mating success: the region between −300 and −91 bp (Figure 7, A and B). This 35 bp shows 100% identity among wild-type inbred laboratory D. melanogaster strains Oregon-R, Canton-S, and Ives (“iso-female line 2”), and also 82 wild-type strains recently collected from Napa Valley, North Carolina, and Zimbabwe (see above). There is 100% identity between D. melanogaster and its sibling species D. simulans, D. mauritiana, and D. sechellia along the first 34 bases, and a shared polymorphism among the three sibling species at base 35 (Figure 7C). In the more diverged species D. subobscura and D. pseudoobscura (∼35 MY), there is also 100% identity along the 34 bp, with the caveat of a 22-bp insertion at the sixth base position in each species (Figure 7C). There are two SNPs between these two species within the 22-bp insertion sequence. Base 35 is identical between D. subobscura and D. pseudoobscura and the three sibling species, but is different from D. melanogaster. Finally, the most diverged species that we investigated are 85.7% (in the case of D. virilis) and 82.9% (D. novamexicana) identical to D. melanogaster and do not share the 22-bp insertion with D. subobscura and D. pseudoobscura (Figure 7C). Both species share base 35 with every species except D. melanogaster, suggesting a recent base change at that site in the D. melanogaster lineage. Additional previously published sequence data from the D. ananassae, D. willistoni, and D. mojavensis genome projects also show perfect or nearly perfect conservation within this 35-bp region, but not outside it (data not shown). This 35-bp sequence may have a conserved cis-regulatory function with respect to male sexual behavior, which the 22-bp inserts in D. subobscura and D. pseudoobscura could modify. No D. melanogaster SNP variation was found within this 35-bp region (see above).

Figure 7.

Molecular evolutionary analysis of the MRS. (A) Map of the yellow locus of D. melanogaster showing the start of transcription at +1 bp, and y mutant landmarks at −300 and −91 bp. Deletion of the region between −300 and −91 bp (solid black MRS line) reduces male mating success to null-allele levels. Insertions within the region between −300 and +1 bp (solid plus dashed MRS line) significantly reduce male mating success to varying degrees. The blue square demarcates a highly conserved 35-bp sequence from −214 to −180 bp. (B) The 105-bp sequence from D. melanogaster and D. virilis, which shows the 35-bp conserved sequence in detail. This sequence is 85.7% identical between the species. The representative 35 bp immediately upstream and downstream of the sequence of interest is poorly conserved with 31.4% and 22.9% identity, respectively. The purple rectangle demarcates a putative Dorsal transcription factor binding sequence. This consensus sequence was identified using the web-based program TFSEARCH, which utilizes the TRANSFAC database (see Thisse et al. 1991; Akiyama 1998; Heinemeyer et al. 1998). The three green rectangles demarcate three putative DSX-binding sites, two of which overlap the putative Dorsal site. Consensus DSX sequences are from Erdman et al. (1996). (C) Alignment of the conserved 35-bp sequence across 12 Drosophila species. The right column indicates percentage identity of sequence. The (*) indicates that the percentage identity calculated does not include the 22-bp insertion in the marked species. Across the Drosophila species that we analyzed, there is only a single base change in the putative Dorsal-protein-binding site, in D. novamexicana. The putative Doublesex-protein-binding sites are perfectly conserved across all the species in our analysis.

Candidate transcription factors for regulation of y via the MRS:

The male-specific sexual differentiation transcription factor FRU^M may regulate y with regard to courtship behavior and MMS (Radovic et al. 2002; Drapeau et al. 2003). This regulation may occur via the MRS, although currently there is no direct evidence for this. In any case, given that temporal, spatial, and sexual regulation of y expression is likely to be more complicated than FRU^M acting in isolation, we used a bioinformatic approach to identify additional transcription factors that might regulate y in the CNS by binding to sequences within the MRS.

The web-based program TFSEARCH allowed us to search the MRS for consensus sequences potentially binding known Drosophila transcription factors in the TRANSFAC database (Akiyama 1998; Heinemeyer et al. 1998). A significant match for the Dorsal (DL) binding sequence (see Thisse et al. 1991) was found within the MRS, a 10/11 nt match to the consensus of (A/C/T)(C/G)(A/C/G)G(A)₄(A/C/T)C(A/C/G), and a 9/11 nt match to the “high-affinity” binding site of GGG(A/T)₅CC(C/A) described by Markstein et al. (2002). Remarkably, this putative DL consensus sequence falls within the highly conserved 35-bp region of the 300-bp MRS (Figure 7, B and C).

Giot et al. (2003) found that Dorsal binds to the Doublesex (DSX) protein in a yeast two-hybrid assay (see also the Fly GRID database located at http://biodata.mshri.on.ca/fly_grid/servlet/searchpage). DSX is a transcription factor that is crucial for proper sexual differentiation, and it has sex-specific products DSX^M and DSX^F that compete for the same binding sites (e.g., Hildreth 1965; Baker and Ridge 1980; Burtis and Baker 1989; Coschigano and Wensink 1993; Zhu et al. 2000). Function of dsx appears to be required mainly for nonbehavioral sexual differentiation outside the CNS (for example, sex-specific pigmentation for which Yellow is necessary and sufficient; Jursnich and Burtis 1993; Kopp et al. 2000), although DSX is found in the CNS at the same stage as Yellow and FRU^M (Lee et al. 2002) and is required for some aspects of normal sexual behavior (Villella and Hall 1996). Because of the in vitro link between DL and DSX, We utilized information about binding sites/affinities for DSX from Burtis et al. (1991) and Erdman et al. (1996) to search the entire 300-bp MRS for potential DSX-binding sites. The only sequence within the MRS to resemble the DSX consensus binding sequence (most important portion underlined) of CTACAAAGT from Burtis et al. (1991) or (G/A)NNAC(A/T)A(T/A)GTNN(C/T) from Erdman et al. (1996) was, again, within the 35-bp highly conserved region (Figure 7, B and C). Three potential DSX-binding sequences are located within the subregion of the 35 bp that is conserved among all species, except a single nucleotide substitution in D. novamexicana. Two of these are exact matches at 6/7 bases, and the third is an exact match but for a 2-nt insert two bases from one end. One of these potential DSX-binding sites is disrupted by the 22-bp inserts in D. subobscura and D. pseudoobscura (Figure 7C). The close proximity of the DL- and DSX-binding sites, combined with the in vitro two-hybrid results of Giot et al. (2003), suggest that DL and DSX might bind to each other and also within the MRS to regulate y transcription.

DISCUSSION

Courtship genetics—major issues:

Animal courtship in nature is of obvious adaptive importance. Laboratory courtship often recapitulates what is found in the wild, and it can be quantified and manipulated (e.g., Hebets and Uetz 1999; Hebets 2003). Therefore its genetic basis can be uncovered. The rationale for examining the genetic basis of instinctual courtship behavior is threefold. First, a fundamental and understudied question of developmental biology is: How does gene action in the immature nervous system build circuitry specifically dedicated to adult behaviors? Second, a major problem in the field of evolutionary neurobiology is: How do genes change the structure and properties of nervous systems, to allow the evolution of complex behaviors? Third, from a quantitative genetic viewpoint, we can ask: Is the genetic architecture of rapidly evolving courtship behavior different from that of other behaviors whose genetic architecture is more slowly evolving (e.g., circadian-regulated activity; Allada et al. 2001; Panda et al. 2002)?

Drosophila courtship can be utilized as a model system with which to answer these questions. Many genes that are required for normal D. melanogaster male courtship have been identified, including two key transcription factors, which are upstream of at least a subset of these genes: fru and dsx (Hall et al. 1982; Hall 1994a; Greenspan 1997; Yamamoto et al. 1997, 1998; Yamamoto and Nakano 1998, 1999; Goodwin 1999; Gaines et al. 2000; Orgad et al. 2000; Romanova et al. 2000; Baker et al. 2001; Greenspan and Ferveur 2000; Sokolowski 2001; Boll and Noll 2002; Dauwalder et al. 2002; Subocheva et al. 2003; Grosjean et al. 2004). Besides fru and dsx, most of these genes have not been well-characterized from the standpoint of their transcriptional regulation, biochemical function, or evolution with regard to male-specific behaviors.

The y gene of D. melanogaster is an excellent candidate for study with regard to understanding the origins of innate sexual behavior in insects. Reasons for this include: (1) Genetic mutations, antibodies, and other sophisticated tools (e.g., the ability to make somatic mosaics via Gal4-mediated UAS misexpression and RNA interference) are available for studying the effects of changing y gene regulation (e.g., Calleja et al. 1996; Piccin et al. 2001; FlyBase 2003); (2) the y gene is under male-specific regulation by the sex determination pathway (Radovic et al. 2002; Drapeau et al. 2003); (3) mutations at y disrupt a specific subset of the behaviors performed during the male courtship ritual (in this case, wing extension) (Bastock 1956; Burnet et al. 1973; Drapeau et al. 2003); and (4) the male sexual behavior function of y appears to be conserved, at least among Drosophila species (Rendel 1944, D. subobscura; Tan 1946, D. pseudoobscura; Frias and Lamborot 1970, D. gaucha; Pruzan-Hotchkiss et al. 1992, D. pseudoobscura; da Silva et al. 2005, D. willistoni ; W. J. Etges, unpublished data, D. mojavensis).

Here, we investigated the regulation of y with regard to male sexual behavior. The research presented is part of a larger investigation of yellow with regard to the three fundamental questions posed above. With two exceptions (Burnet and Wilson 1980; Robertson 1982), the many studies involving the requirement of y for normal D. melanogaster male mating success levels have exclusively involved the y¹ allele or, in a minority of cases, undesignated pigmentation-null (“type-1,” e.g., Green 1961; Chia et al. 1986; Lindsley and Zimm 1992) alleles of y (Sturtevant 1915; Spett 1931; Diederich 1941; Merrell 1949; Mayr 1950; Meyers 1953; Bastock 1956; Barker 1962; Schroeck 1971; Burnet et al. 1973; Threlkeld et al. 1974; Dow 1975, 1977a,b,c,d; Tracey and Espinet 1976, 1977; Wilson et al. 1976; Oakeshott and Hayman 1979; Bradman et al. 1981; Mizuguchi and de Almeida 1983; Heisler 1984; Hamerlynck 1994; Drapeau et al. 2003). In this article, we extended these earlier results in numerous ways.

The y¹ allele is a null with regard to male mating success:

We unambiguously showed that y¹ is a null allele for male mating success. Our consideration of this point is nontrivial, since reputable literature discussion of y function with respect to sexual behavior (e.g., Hall 1994a, 2002) is based on the studies cited above that were conducted with y¹. This mutant allele has been previously characterized as an A-to-C transversion in the first translated codon, resulting in an altered mRNA product (Geyer et al. 1990; Morris et al. 1999). Therefore, before this study, with respect to male sexual behavior the hemizygous y¹ allele could possibly have been a null allele, a hypomorph, or even a neomorph, if the mutated mRNA had a unique behavioral function unrelated to the normal purpose of the y gene (Muller 1932). Our data (Table 2, Figure 2) suggest that the mating success phenotype of y¹ is equivalent to that of males carrying either of two null alleles caused by the deletions Df(1)y-ac²² and y^59b (Table 1). We have also shown that y¹ has a more severe MMS phenotype than two mild hypomorphs for mating behavior, y^a77 and y^2S (Figures 2 and 3, Table 2). These behavior–genetic data, combined with immunocytochemical evidence that a new anti-Yellow antibody does not recognize any proteins in either y¹ or Df(1)y-ac²² flies (Figure 5, B and E; see also Drapeau 2001b; Radovic et al. 2002; Wittkopp et al. 2002a; Drapeau et al. 2003), fulfill the requirements of a null hemizygous allele classification for y¹ with regard to male mating success (Muller 1932).

A regulatory sequence controlling male sexual behavior:

We used y mutant alleles with known molecular lesions as a tool for mapping the regulatory sequence required for normal MMS (Figure 4). In addition to the behavioral data collected from males carrying y null alleles, the utilization of a y⁺ minichromosome to control genetic and environmental backgrounds (Figure 1), combined with the fact that four hypomorphic y alleles with independent origins all significantly reduced male mating success (Table 2, Figure 3), is strong evidence for an overall requirement for y in the specification of adult male sexual behavior. Following the classic example of the per^s, per^L, and per⁰ alleles shortening, lengthening, and abolishing behavioral circadian rhythms, respectively (Konopka and Benzer 1971), some of the most convincing evidence that a gene is directly involved in the control of a behavior is the detection of a range of phenotypic effects of different alleles at the locus. We found statistically significant variation in MMS among the four hypomorphic y alleles that each individually decreased MMS, y⁻⁹¹, y^3d, y^a77, and y^2S. In addition, the nonsignificant MMS results for the remaining eight y hypomorphic alleles with mild-to-severe effects on pigmentation imply that the positive results that we found were not spurious. Our molecular and behavioral data demarcate a region of ∼300 bp whose function is required for normal levels of male mating success.

Two past studies (Burnet and Wilson 1980; Robertson 1982) reported data from a small number of non-null alleles of y whose molecular underpinnings were not known at the time. These two studies reported data collected using a wide variety of experimental designs, varying the number of days the flies spent courting and mating, the size of the mating chamber and whether or not food was present in the chamber, the number of males and females simultaneously in the chamber, the crossing schemes to generate mutant and control males, and the level of inbreeding of the assayed flies. These varied designs yielded a spectrum of results. For example, both studies investigated the y² mutant, which we found to have no effect on male mating success (Table 2, Figure 3). Across all experiments performed by Burnet and Wilson (1980) and Robertson (1982), they measured MMS phenotypes ranging from a strong disadvantage to a slight advantage of y² relative to controls.

In this study, using a standardized, outbred genetic background across all mutants and all controls (Figure 1) and quantifying behavior of individual F₁ mutant or control males with single females (“single-pair matings”) simultaneously in the same mating apparatus with moderate-sized 25-mm-diameter × 20-mm-deep behavioral chambers (Drapeau and Long 2000) allowed us to accurately and sensitively measure the effects of different y mutations on MMS. The minichromosome-rescue strategy employed in our study exquisitely controls genetic and environmental backgrounds between mutant and control subjects, allowing confident measurement of behavior in fully outbred flies. This strategy or similar ones should be useful in behavior–genetic studies on other loci.

We hypothesized that regulatory variants of y that reduced MMS would define a small region of the locus whose function is required for a complex, sex-limited adult behavioral performance. The behavior–genetic data that we collected define an ∼300-bp region upstream of the y transcription start site that contains a regulatory sequence required for normal MMS, the sequence we termed the MRS. Conservatively, the 5′ limit of the MRS was defined by the y^−300gin mutation that carried the gypsy insulator at −300 bp relative to y transcription start and the 3′ limit by the y^a77 insertion at −4 bp (Figure 4). It is likely that the 209-bp sequence between −300 and −91 bp contains the sequence(s) required for MMS, since males carrying the y⁻⁹¹ deletion mutant have severely decreased mating success, while y^−300gin mutant males do not (Table 2, Figures 3 and 4). We propose that the mutants with insertions that are located in the 300-bp region upstream of the transcription start, y^3d, y^a77, and y^2S, reduce or eliminate the function of the 209-bp element, which in turn lowers male mating success to varying degrees (Table 2, Figures 3 and 4).

While this is the most parsimonious interpretation of the data, the true regulation of y with regard to MMS may be more complicated. The y^2S allele, for example, causes a slight reduction in MMS, and experiment-wide significance was detectable only with a large sample size (Table 2, Figure 3). Within trials, effects of this mutant allele were sometimes significantly different and sometimes not distinct from the control males (data not shown). The underlying cause of this variability is currently unclear. It is possible that the y^2S insertion stochastically recruits factors that can substitute for the MRS. Another complicated interpretation is data collected from males carrying the y⁻⁸⁹³ allele. These males had a nonsignificant difference in their overall MMS, but the percentage of success measured for this genotype was similar to that of y^2S males (45.3% vs. 44.9%). The overall sample sizes for these two genotypes also differed (123 vs. 335), giving us more statistical power to detect a difference in y^2S than in y⁻⁸⁹³ (Table 2). Our current interpretation is guided by the fact that four other mutant alleles affecting the same region of the y locus as y⁻⁸⁹³, upstream of −300 bp, did not cause a significant reduction of MMS, and likewise, three mutants affecting the same region as y^2S, −300 to 0 bp, significantly reduced MMS (Table 2, Figures 3 and 4). The finding that y⁻⁸⁹³ flies do not have Yellow in their third instar CNS (Figure 5F, compare to Figure 5, A and B) further clouds our interpretation. It is possible that the MMS and Yellow protein reduction in y⁻⁸⁹³ flies is not caused by an intrinsic regulatory defect at y, but rather because of a position effect due to insertion of the transgene at an ectopic site. Alternatively, it may be the case that while these Yellow⁺ third instar neural cells are associated with adult MMS, they are part of a more complex story. Only a careful series of behavior–genetic experiments with Gal4-mediated somatic neural y⁺ mosaic flies, complementing data in this study, will shed light on these issues.

To put our mapping resolution (i.e., a 209-bp cis-regulatory region required for normal behavior) into perspective, earlier investigations of sequences within the y locus controlling pigmentation defined these regulatory regions as ranging in size from ∼130 to 1200 bp (Geyer and Corces 1987, 1992; Martin et al. 1989). One recent study of D. melanogaster courtship genetics had a focus similar to ours. Boll and Noll (2002) mapped two adjacent cis-regulatory regions necessary for normal male mating speed/time to copulation within a Pox neuro gene to one region of ∼2 kb and one of ∼10 kb. This behavioral phenotype appears to be correlated with expression of the Pox neuro transcription factor in the adult brain. It is not presently clear what the relationship is, if any, between Pox neuro and other courtship genes such as fru and y.

Within the broadly defined 300-bp MRS, we identified four SNP variants, one common and three rare, among a group of 85 wild-type D. melanogaster strains. These SNPs may contribute to naturally occurring variation in male sexual behavior. However, unlike the mutants in Table 1 that had severe (or no) effects on MMS, any effect of these SNP variants on male sexual behavior is expected to be minor. Such small effects can be difficult to detect because of the greater behavioral scrutiny required to find slight differences between mutants and controls, the large sample sizes needed to detect small phenotypic effects, and genetic epistasis and environmental influences that can obscure the primary effect of the y mutation. Furthermore, because the y locus is on the X chromosome, the SNPs cannot be used in a simple quantitative complementation test (Long et al. 1996). Alternatively, these SNPs may be neutral mutations with respect to male sexual behavior.

Adult male pigmentation and behavior are not correlated:

Another way in which we have extended previous work on the y gene and male sexual behavior is by demonstrating that there is no obvious relationship between two adult phenotypes controlled by y. We found that cuticle pigmentation in various adult tissues was not predictive of y mutant male mating success (Table 3). This is an interesting negative result, because there are numerous mechanisms by which pigment/color alteration can cause differences in male mating success (e.g., Burnet and Connolly 1974; Singh and Chatterjee 1987). In total, our data indicate that the adult pigmentation and male sexual behavior phenotypes are developmentally regulated independently. Indeed, the separate, modular control of y transcription with regard to adult pigmentation (e.g., Geyer and Corces 1987; Martin et al. 1989; Gompel et al. 2005) and adult behavior (this study) suggests that individual cis-regulatory modules within a pleiotropic gene provide a mechanism by which multiple evolutionarily important traits can separately evolve.

Molecular function of the mating-success regulatory sequence:

The data presented in this report provide a starting point for understanding the transcriptional regulation of the y gene with respect to male sexual behavior. The MRS now defined, we ask, what is its function? An important clue comes from previous studies of cis-regulatory sequences of y-controlling pigmentation. The sequence between −300 and −91 bp is required for normal levels of larval pigmentation (Geyer and Corces 1987; Martin et al. 1989), implying that this sequence is required for upregulating y transcription in specific larval tissues that normally are pigmented (e.g., Walter et al. 1991). Larval pigmentation is unlikely to be behaviorally relevant in the adult male, since these pigmented structures disappear during metamorphosis. (Furthermore, y¹ larvae act normally in at least one larval behavioral paradigm; see Mazzoni et al. 2005.) However, at a molecular level, the 209-bp region probably contains enhancer sequences that bind transcription factors required for normal development of both larval pigmentation and adult male behavior. Enhancers are (1) often identified with selective gene expression in particular tissues, (2) distance independent, and (3) orientation independent. While the MRS/larval pigment control region DNA appears to direct y expression in specific tissues, distance independence and orientation independence remain to be tested for this y regulatory region. Following the approach of Geyer and Corces (1987), future work can focus on determining whether the MRS region functions as a transcriptional enhancer.

Candidate transcription factors for regulation of y via the MRS:

Our working model is that the MRS binds factors that activate y transcription in CNS cells relevant to male sexual behavior (Radovic et al. 2002; Drapeau et al. 2003). One factor that might bind within the MRS is the male-specific FRU isoform BM, notated FRU^M{BM}, which we have shown by temporal, spatial, and sexual misexpression to be sufficient for Yellow accumulation, as detected by immunohistochemistry and Western blotting (Drapeau et al. 2003). Experiments with fru¹, fru³, and fru⁴ mutants suggest that the zinc-finger regions of FRU are required for Yellow production in these cells (Drapeau et al. 2003). On the basis of the results reported here, we hypothesize that FRU^M{BM} or a downstream target of FRU^M{BM} binds within the MRS to activate y transcription in larval neural cells crucial for the development of adult male behaviors.

The specific function of the 60-kDa Yellow protein is unknown, but there are hypotheses about its requirement in pigmentation and behavior. Yellow is secreted from the cells where the y gene is expressed (Hannah 1953; Geyer et al. 1986; Kornezos and Chia 1992; Radovic et al. 2002; Wittkopp et al. 2002a; Drapeau et al. 2003). Assuming that Yellow is diffusable, it may then influence the properties of nearby cells in the CNS via a hormone- or growth-factor-like mechanism (Drapeau 2003). An alternative but nonmutually exclusive hypothesis is that Yellow crosslinks dopamine derivatives during melanization (Geyer et al. 1986). This function requires the amino acid cysteine (Cys), and indeed Yellow is Cys-rich, unlike other cuticle proteins. If Yellow has this function in the CNS, it could exert its effect through a melanin-signaling pathway. Either of these proposed Yellow functions could be relevant to the construction of a neural circuit subserving male courtship behavior (Baker et al. 2001; Stockinger et al. 2005).

In this report, we presented evidence that the transcription factors Dl and DSX may bind within a highly conserved 35-bp region within the MRS. DSX is an excellent candidate to be a direct regulator of y expression. First, as previously mentioned, it is a member of the sex determination cascade and is expressed in the CNS, and mutations in dsx alter sex-specific behaviors. Second, the normal function of the takeout gene with regard to normal male courtship is dependent on the function of both fru and dsx (Dauwalder et al. 2002), and therefore this may reasonably apply to y as well. (We did not investigate regulation by dsx during our previous work with fru and y.) Third, the sex-specific DSX^M and DSX^F proteins that share binding sites suggest a mechanism by which y would be upregulated in males and suppressed in females (see, for example, Waterbury et al. 1999; Dauwalder et al. 2002). While this discussion could stem from mere coincidence, our bioinformatic data provide fruitful avenues for future molecular-genetic and biochemical research on the MRS. We speculate that the well-studied transcription factors Dorsal and Doublesex bind within a small, highly conserved sequence in the y locus to, in collaboration with FRU^M{BM}, control behavioral phenotypes.

Functional analysis of the MRS:

Using genetic and biochemical methods, we presented evidence that the MRS directs expression of y in a small number of cells in the third instar CNS and that Yellow protein in these cells is associated with normal MMS. On the basis of this, we hypothesize that the presence of Yellow in these third instar neural cells is required for the development of normal male courtship behavior (specifically, WE), with normal mating success as a consequence. Our earlier demonstration that a global third instar CNS Gal4 driver combined with UAS-y⁺ in a y¹ background is sufficient to rescue low levels of WE and MMS supports this idea (Drapeau et al. 2003). However, with our current data we cannot exclude that y expression in the CNS at other stages of development may play a role, nor can we exclude that y expression in other tissues is relevant, with the exception of the adult body cuticle (see results).

The next step in this line of research is to study the MRS in isolation and understand its properties with regard to transcriptional regulation. In this work, we look to the case of the “E-box” sequence and its role in the generation of circadian molecular rhythms and overt rhythmic behavior. A 69-bp regulatory sequence was mapped, then shown to act as an enhancer of period transcription, and subsequently dissected further using molecular and biochemical techniques to demonstrate that the shorter E-box sequence within the original 69 bp was sufficient to generate rhythms (e.g., Hao et al. 1997, 1999; Darlington et al. 1998). Future work on the MRS within the y locus should shed light on the molecular-genetic mechanisms underlying a complex adult instinct with relevance to reproductive isolating mechanisms and speciation.