Molecular Genetic Dissection of the Sex-Specific and Vital Functions of the Drosophila melanogaster Sex Determination Gene fruitless

Corresponding author: Bruce S. Baker, Department of Biological Sciences, Stanford University, Stanford, CA 94305., bbaker{at}cmgm.stanford.edu (E-mail)

Communicating editor: R. S. HAWLEY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A multibranched hierarchy of regulatory genes controls all aspects of somatic sexual development in Drosophila melanogaster. One branch of this hierarchy is headed by the fruitless (fru) gene and functions in the central nervous system, where it is necessary for male courtship behavior as well as the differentiation of a male-specific abdominal structure, the muscle of Lawrence (MOL). A preliminary investigation of several of the mutations described here showed that the fru gene also has a sex-nonspecific vital function. The fru gene produces a complex set of transcripts through the use of four promoters and alternative splicing. Only the primary transcripts produced from the most distal (P1) promoter are sex-specifically spliced under direction of the sex-determination hierarchy. We have analyzed eight new fru mutations, created by X-ray mutagenesis and P-element excision, to try to gain insight into the relationship of specific transcript classes to specific fru functions. Males that lack the P1-derived fru transcripts show a complete absence of sexual behavior, but no other defects besides the loss of the MOL. Both males and females that have reduced levels of transcripts from the P3 promoter develop into adults but frequently die after failing to eclose. Analysis of the morphology and behavior of adult escapers showed that P3-encoded functions are required for the proper differentiation and eversion of imaginal discs. Furthermore, the reduction in the size of the neuromuscular junctions on abdominal muscles in these animals suggests that one of fru's sex-nonspecific functions involves general aspects of neuronal differentiation. In mutants that lack all fru transcripts as well as a small number of adjacent genes, animals die at an early pupal stage, indicating that fru's function is required only during late development. Thus, fru functions both in the sex-determination regulatory hierarchy to control male sexual behavior through sex-specific transcripts and sex-nonspecifically to control the development of imaginal discs and motorneuronal synapses during adult development through sex-nonspecific transcript classes.

MULTIPLE promoters, alternative splicing, and complex sets of cis-acting sequences governing temporal and spatial patterns of gene expression are all employed to allow single genes to carry out multiple unrelated biological functions. The fruitless (fru) gene of Drosophila melanogaster is a particularly interesting gene that is representative of such multifunctional genes. fru has been shown to function as a member of the Drosophila sex determination regulatory hierarchy (RYNER et al. 1996 ; HEINRICHS et al. 1998 ), where it acts to build the potential for male sexual behaviors into the male central nervous system (ITO et al. 1996 ; RYNER et al. 1996 ; VILLELLA et al. 1997 ; GOODWIN et al. 2000 ; LEE et al. 2000 ; USUI-AOKI et al. 2000 ). In addition, fru function is essential for the viability of both sexes (RYNER et al. 1996 ).

The multiple functions of the fru gene are encoded by a complex set of sex-specific and sex-nonspecific transcripts that are generated by the use of four promoters (P1–4) and alternative splicing at both the 5' and 3' ends of the primary transcripts (ITO et al. 1996 ; RYNER et al. 1996 ; GOODWIN et al. 2000 ; USUI-AOKI et al. 2000 ; L. C. RYNER, S. F. GOODWIN, T. CARLO, M. FOSS, J. C. HALL, B. J. TAYLOR and B. S. BAKER, unpublished results). The transcripts from each of fru's promoters have long open reading frames that encode closely related proteins belonging to the BTB-Znf family (cf. ALBAGLI et al. 1995 ; HU et al. 1995 ). Transcripts generated from the distal (P1) fru promoter are spliced sex-specifically under the control of transformer (TRA) and transformer-2 (TRA-2) proteins in females, whereas in males a default splice occurs (RYNER et al. 1996 ; HEINRICHS et al. 1998 ). These sex-specifically spliced P1 transcripts are expressed in a small subset of neurons in the central nervous system (CNS; RYNER et al. 1996 ; LEE et al. 2000 ; USUI-AOKI et al. 2000 ) and are likely responsible for fru's role in male sexual behavior. In females, the splicing of the P1-derived transcripts gives rise to mRNAs with the potential to encode proteins with a BTB domain at their amino termini and one of three alternative Zn finger pairs at their carboxy termini (GOODWIN et al. 2000 ; L. C. RYNER, S. F. GOODWIN, T. CARLO, M. FOSS, J. C. HALL, B. J. TAYLOR and B. S. BAKER, unpublished results). In males, the default splicing of the P1 transcripts generates male-specific mRNAs, encoding proteins that differ from the proteins predicted to be produced from the female mRNAs by the addition of 101 amino acids N-terminal to the BTB domain. Strikingly, immunohistochemistry with anti-FRU antibodies shows that the mRNAs produced from the P1 fru promoter are not translated in females, whereas they are translated in males (LEE et al. 2000 ; USUI-AOKI et al. 2000 ). The transcripts from the three more proximal promoters (P2, P3, and P4) encode sex-nonspecific proteins that differ from the male-specific proteins by having just short stretches of amino acids preceding the BTB domain (L. C. RYNER, S. F. GOODWIN, T. CARLO, M. FOSS, J. C. HALL, B. J. TAYLOR and B. S. BAKER, unpublished results). The roles of these sex-nonspecific proteins are not known, but it has been suggested that one or more of them are responsible for fru's vital function (RYNER et al. 1996 ). Moreover, expression data show that one or more of these products are expressed widely in the nervous system as well as in some other tissues in both sexes (RYNER et al. 1996 ; LEE et al. 2000 ).

Most of the fru mutations or mutant combinations studied to date have been viable hypomorphs (ITO et al. 1996 ; RYNER et al. 1996 ; VILLELLA et al. 1997 ; GOODWIN et al. 2000 ). Nevertheless, these studies have revealed that wild-type fru function is required for most steps in male courtship behavior but has no detectable role in female courtship behavior. fru males typically display reduced levels of courtship to females, as measured by the amount of time that a male courts a prospective mate (HALL 1978 ; GAILEY and HALL 1989 ; GAILEY et al. 1991 ; RYNER et al. 1996 ; VILLELLA et al. 1997 ; GOODWIN et al. 2000 ). In addition, males of many fru genotypes are defective in specific parts of the courtship sequence (GAILEY et al. 1991 ; RYNER et al. 1996 ; VILLELLA et al. 1997 ; GOODWIN et al. 2000 ). For example, wild-type males choose females rather than males as the appropriate sexual partner, while fru males fail to distinguish between the sexes and as a consequence court females and males roughly equally (RYNER et al. 1996 ; VILLELLA et al. 1997 ; GOODWIN et al. 2000 ). Indeed, when fru males are housed together, male-male courtship leads to the formation of groups of courting males (HALL 1978 ; GAILEY and HALL 1989 ; GAILEY et al. 1991 ; ITO et al. 1996 ; RYNER et al. 1996 ; VILLELLA et al. 1997 ; GOODWIN et al. 2000 ). At a slightly later stage of courtship, when wild-type males extend a wing and vibrate it to produce a courtship song, certain fru mutants—for example, fru³, fru⁴, and fru^sat males—rarely extend a wing toward females and do not produce courtship pulse song during the short bouts of wing extension that they produce (RYNER et al. 1996 ; VILLELLA et al. 1997 ). Copulation is not attempted by males of most fru genotypes, rendering these males sterile (but see GAILEY et al. 1991 ; VILLELLA et al. 1997 ; GOODWIN et al. 2000 ). Finally, it was recently found by examining fru mutant combinations that do copulate that behaviors during mating, such as transfer of seminal fluids and sperm and the duration of mating, are also governed by fru (LEE et al. 2001 ). While the fru mutations characterized in these studies have strong effects on male courtship behaviors, the full phenotypic consequence of removing fru function has not been established.

To better understand fru's functions, we conducted mutageneses of existing fru P-element alleles with the hope of identifying loss-of-function fru alleles and alleles that removed subsets of fru's transcript classes. Eight new lesions were obtained in the fru locus. Our initial analysis of some of these mutants (RYNER et al. 1996 ) showed that fru's role in controlling male-specific behavior was more extensive than had been previously recognized and that fru had a vital function in both sexes (see also ITO et al. 1996 ). From analyzing male courtship behaviors in numerous mutant combinations, we show here that fru is involved in every step of the courtship ritual, thus extending the role of fru to encompass all aspects of male courtship. We analyzed in detail the anatomical, neuronal, and behavioral phenotypes associated with these new lethal fru lesions and combinations of these lesions with preexisting mutations and deficiencies of fru. Among the new anatomical phenotypes seen in lethal fru mutants were defects in the adult derivatives of the imaginal discs and a reduction in the terminal arborization of motor axons on adult abdominal muscles, suggesting that fru plays a role in the differentiation of the adult epidermis and CNS. In addition, we determined the effects of the new mutations on the expression of the different fru transcripts and correlated fru's sex-specific and sex-nonspecific functions with the expression of particular classes of transcripts. Our results establish that fru functions sex-specifically in the sex-determination regulatory hierarchy to control male sexual behavior and sex-nonspecifically to control the development of imaginal discs and motorneuronal synapses during adult development.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Generation of new fru mutations:
Fly strains and crosses: Flies were reared at 25° in 70% relative humidity on a standard cornmeal, yeast, molasses (or dextrose), and agar diet with proprionic acid and/or tegosept added to inhibit mold.

Previously outcrossed stocks of fru^1-4 (for details see GOODWIN et al. 2000 ) were balanced with TM3, Sb ry, MKRS, Sb ry, TM6C, Sb Tb, or TM6B, Hu Tb e balancer chromosomes (for details on these markers, see LINDSLEY and ZIMM 1992 ). The fru² allele was recently reextracted by crossing to a "Cantonized" white stock. The fru^sat and Df(3R)fru^sat15 alleles (ITO et al. 1996 ; kindly provided by D. Yamamoto) were maintained over TM6B and TM6C balancer chromosomes, respectively. Crosses with third chromosomal deficiencies that have break-points in, or close to, the fru locus, Df(3R)Cha^M5, Df(3R)Cha^M7, Df(3R)P14, Df(3R)BX5, Df(3R)gl^BX7, and Df(3R)148.5-1 (Table 1 and Fig 1A; GAILEY and HALL 1989 ; GAILEY et al. 1991 ), were used in conjunction with the newly generated fru mutations to analyze mutant phenotypes.

View larger version (25K): In this window In a new window Download PPT slide   Figure 1. Genetic and molecular map of the fru locus. (A) The cytological map of the 90–91 region (right arm of the third chromosome). For deficiencies, the deleted regions are indicated by thick solid lines (see Table 1 for details of the mutations). The locations of the chromosomal breaks in fruw9, fruw12, and fruw27 alleles are denoted by the arrow. (B) The molecular map of the fru locus. The insertion sites of the fru2, fru3, fru4, and frusat P elements and the breakpoints of the fruw9, fruw12, and fruw27 chromosomal aberration mutations are shown at a higher level of resolution. The ends of relevant deficiencies are shown, with the thick solid lines demarcating missing DNA and the dashed lines representing the restriction fragments to which the breaks were mapped. The positions of the four promoters (P1–P4) are mapped to their relevant restriction fragments of the fru genomic DNA (adapted from RYNER et al. 1996 ). Each alternative 3' end encodes a pair of Zn-f domains and is labeled A, B, or C according to its distal-to-proximal location in the gene. The P1 transcripts are sex-specifically spliced by the action of the TRA and TRA2 proteins interacting with 13-nucleotide (nt) repeats, resulting in sex-specific transcripts that produce proteins differing at their amino termini. By Northern analysis, three P1 transcript classes have been detected in male and female heads that correspond to usage of each of the alternative 3' ends (GOODWIN et al. 2000 ). The full transcript complexity is not known for transcripts from the other promoters and so the 3' ends for P2, P3, and P4 are not included. Note that the orientation of the molecular map is reversed with respect to the cytological map.

Mutageneses: To generate more severely mutant derivatives of the fru² allele, which is due to the insertion of a P element (p^p[wA]) in the fru locus, fru² homozygous males were X rayed (4000 R, 115 V, 5 mA, 1-mm Plexiglas filter, Torrex X-ray machine) and crossed to w; Dr Pr/TM3, Sb ry virgin females. From 62,000 progeny, 31 white-eyed flies that had lost the w⁺ marker of the P element in fru were obtained. A total of 16 lines were established by backcrossing three times to w; Dr Pr/TM3, Sb ry flies; lines were not established from the other 15 animals. These 16 lines were tested for sterility over the fru² parental chromosome. Seven new fru alleles, fru^w9, fru^w10, fru^w12, fru^w13, fru^w24, fru^w25, and fru^w27, were identified on the basis of the fact that these chromosomes were weakly fertile or sterile over fru² (Table 4). The remaining nine chromosomes were fertile over fru², cytologically normal in the fru region, and were not studied further.

Excisions of the fru⁴ P element (P[lacZ; ry⁺]; CASTRILLON et al. 1993 ) were generated using a standard transposase source (ROBERTSON et al. 1988 ). Of the 90 ry^- lines generated from the remobilization of the fru⁴ P element, Southern analysis showed that 37 were precise, or nearly precise, excisions of the P element, 23 were small deletions, and 3 were lethal. One of the lethal mutations, Df(3R)fru^4-40, which is lethal over Df(3R)Cha^M5 but not over Df(3R)P14 (Table 3), was used in this study.

Molecular characterization of the new fru alleles:
Determining the nature of their chromosomal rearrangements: To determine the type of lesion present in the new fru mutants, cytological analysis was done on salivary chromosomes from fru^-/+ third instar larvae.

Southern analyses were carried out on fru^w9/TM3, fru^w12/TM3, and fru^w27/TM3 adults to localize the position of these chromosomal breakpoints. We used labeled P^p[wA] DNA (HAZELRIGG et al. 1984 ; Flybase segment ID no. FBmc0000129) to probe Southern blots of restriction digests of genomic DNAs of the heterozygous fru mutants and control genomic DNAs obtained from flies of the parental stock used in the mutagenesis, fru²/TM3, and flies heterozygous for a complete deficiency of the region, Df(3R)fru^w24/TM3.

Df(3R)fru^4-40 was mapped using quantitative Southern blots of genomic DNA from mutant heterozygotes, wild-type control, homozygous fru⁴ control, and balancer control flies probed with a series of phage probes (f3A, f2A, f1H, f4B, and f5C) spanning the fru locus (see Fig 1A, RYNER et al. 1996 ). These experiments showed that Df(3R)fru^4-40 has one end in the fru⁴ P element and extends distally beyond the P1 promoter.

Df(3R)fru ^sat15 was roughly mapped by a combination of Southern blot analysis and genomic PCR using oligo primers derived from plasmid subclones from across fru. To determine the exact endpoints of the deficiency, oligonucleotides flanking the breakpoints were utilized with PCR to amplify a fragment containing the sequences flanking the breakpoints. Sequencing of this product and comparison to the genomic sequence for this region showed that the deletion begins 20 nucleotides into the inverted repeat of the fru^sat P element and deletes all nucleotides from 96,551 to 13,839 of ACC no. 003722, deleting the entire coding sequence of fru (data not shown). In addition, Df(3R)fru^sat15 completely deletes the four transcription units immediately proximal to fru as well as parts of two overlapping transcription units that are the fifth and sixth transcription units proximal to fru.

Transcript analysis by reverse transcription PCR: To determine which fru transcripts were produced by various fru alleles, the following genotypes were analyzed: Df(3R)fru^w24/T(3;het)fru^w12, Df(3R)fru^w24/In(3R)fru^w27, T(3;het)fru^w12/In(3R)fru^w27, Df(3R)fru^4-40/Df(3R)fru^sat15, and Df(3R)Cha^M5/ T(3;het)fru^w12. Since most of these genotypes die before adulthood, we collected 50 fru mutant animals prior to their lethal phase, which was 1- to 2-day-old pupae for Df(3R)fru^w24/T(3;het)fru^w12, Df(3R)fru^w24/In(3R)fru^w27, T(3;het)fru^w12//In(3R)fru^w27 and adult flies for Df(3R)fru^4-40/Df(3R)fru^sat15 and Df(3R)Cha^M5/T(3;het)fru^w12. Each of these genotypes was generated from crosses using stocks balanced with the TM6B, Hu Tb e balancer. Animals were quick frozen and stored at -80°. Total RNA was extracted using the TRIzol reagent (GIBCO-BRL, Grand Island, NY). Reverse transcription was performed on total RNA from an equivalent of 2.5 animals for a given genotype using a primer complementary to a sequence from a region common to all known fru transcripts [primer (fru-8-rev) 5' gtgagaccacgcacctgtgcag-3']. One-tenth of the reverse transcription reactions were used for the PCRs. Six different primer sets were used for the PCR to generate products diagnostic of the following: (A) transcripts containing the common coding portion of fru [primers 5'-aacactgaccaaggagcgatg-3' (fru-25) and 5'-atgggcagcgaactctggcc-3' (fru-26-rev)]; (B) transcripts from the P1 promoter spliced in the female pattern [first round primers 5'-ccagatcgaaagagaatatcatca-3' (fru-2) and 5'-cagcgcaagcagaattgctgg-3' (fru-7-rev); second round primers 5'-taattctaaccgaaagtaagcatag-3' (fru-32) and fru-7-rev]; (C) fru transcripts from the P1 promoter spliced in the male pattern [primers 5'-cttccgcccgcatcccctag-3' (fru-31) and fru-26-rev]; note that this primer pair also amplifies the female-specific product from this region; (D) transcripts initiated from the P2 promoter [primers: both rounds forward primer 5'-atcataaaatcgctcggttttagtt-3' (fru-29); first round reverse primer fru-26-rev; second round reverse primer 5'-catgaactcgagcagagatcgca-3' (fru-55-rev-Xho)]; (E) transcripts initiated from the P3 promoter [first round forward primer 5'-ctgagaacgtgcgcgagtgtt-3' (fru-11); second round forward primer 5'-caaagtgagtgagatacaatcgc-3' (fru-12); reverse primer used in both rounds fru-7-rev]; (F) transcripts initiated from the proximal-most fru promoter, P4 [primers 5'-ttacactaactattggctgctgg-3' (fru-40) and fru-26-rev; (G) an additional set of primers that amplify the male P1 product was used in the case of the Cha^M5/fru^w12 mutant combination: forward primer 5'-gcattacgcggccttggact-3' (fru-28) and reverse primer fru-26-rev. PCR products were size fractionated on agarose gels and analyzed on Southern blots to confirm that they were bona fide fru products. Note that each of the primer sets used amplifies a product that contains at least one splice junction so that products generated from contaminating genomic DNA, if any, are distinguishable from those generated from fru transcripts. The identity of the PCR products detected in these experiments was confirmed in all cases by probing these Southern blots with end-labeled oligonucleotide probes homologous to fru mRNA sequences expected to be amplified by each primer pair or with probes that span at least part of the expected fru products. These oligo probes used did not overlap the primer sequences used for amplification.

In situ hybridization to fru mutant animals: In situ hybridizations to fru mutant animals were carried out as described in GOODWIN et al. 2000 on 20-µm horizontal cryostat sections. Single-stranded riboprobes were synthesized from fragments of fru cDNAs subcloned into Bluescript pSK(+). The labeling reaction used T3 or T7 polymerase with digoxygenin-coupled nucleotides according to manufacturer's instructions (DIG RNA labeling kit; Boehringer Mannheim, Indianapolis). Yields were estimated by dot blot of serially diluted labeled probes by comparison with prelabeled RNA standards (Boehringer Mannheim). The P1 promoter probe (Probe S; RYNER et al. 1996 ) included nucleotides 160,236–159,918 of ACC no. AE003722, plus nucleotides 1–261 of ACC no. U27492. The common coding region probe (Probe C; RYNER et al. 1996 ) consisted of nucleotides 2785–3612 in ACC no. U72492.

Behavioral analysis of fru mutants:
Adult fru mutant males were tested for a variety of courtship and noncourtship behaviors. To test for sterility, males were collected at eclosion and stored either individually or as a group of 8–10 males per vial. After aging for 3–4 days, single males were placed with two to three wild-type virgin females in food vials and the presence or absence of the progeny was scored 7 days later (GAILEY and HALL 1989 ; GAILEY et al. 1991 ; VILLELLA et al. 1997 ). Vials in which the males had died were not counted.

Males to be tested for their courtship behavior were collected just after eclosion and aged individually for 6–10 days. The test male was placed in a small observation chamber with either another male of the same genotype or a virgin Canton-S female. The pair was then video recorded for 5–8 min (cf. VILLELLA et al. 1997 ). The percentage of time that a male spends courting [courtship index (CI)] another male (CI m m) or female fly (CI m f) during the observation period was determined by viewing the video tapes and logging the behaviors with a digital event recorder, leading to elementary computations of accumulated-time-logged/total-observation-time for a given playback (VILLELLA et al. 1997 ). When the male is paired with another male, the CI is calculated only for the first male to initiate courtship toward the other male. When the male is paired with a female, the CI represents the time that the male courts the female. The wing extension index (WEI; VILLELLA et al. 1997 ), which is the amount of time that the male extended his wing during the observation period, was also calculated. In addition, courtship song was simultaneously recorded onto the videotapes, and then a subset of the m f recording was analyzed for various song parameters using LifeSong (VILLELLA et al. 1997 ).

Observations of male courtship behavior in all-male groups were used to assess the degree that mutant males courted each other. Males were collected at eclosion and aged individually for 5–6 days. After grouping eight males of the same genotype together in a food vial for 3–4 days, an observation of multi-male courtship interactions for each genotype was made in the late afternoon or early evening. Animals were housed and observations were made under standard conditions of 25° and 70% humidity. A courtship chain was defined as an episode during which three or more of the eight males courted one another. The percentage of time that three or more males courted in such group interactions during a 10-min observation period was logged using a timer and computed as the chaining index (ChI; VILLELLA et al. 1997 ).

General activity measurements: To measure short-term locomotor activity, males were collected at eclosion and aged individually for 7–9 days (KULKARNI and HALL 1987 ). Single males were then placed in a small plexiglass chamber (1 cm diameter x 6 mm height) in which a filter paper with a single line dividing the chamber into two equal halves was placed at the bottom. After a 2- to 3-min accommodation period (which started just after introducing males into the chambers) the number of times each male crossed the center line in a 3-min observation period was counted with a hand-held counter. These observations were done in the late afternoon, usually 2–4 hr before the lights go off.

Statistics: CIs and ChIs were subjected to arcsine square root transformations (cf. VILLELLA et al. 1997 ) and then the studentized residuals were tested for normal distribution approximations (SOKAL and ROHLF 1995 ). Short-term activity and courtship song (interpulse intervals) data were not transformed. One-way ANOVAs, followed by subsequent planned pairwise comparisons, were performed on all behavioral and song data (see table legends for further statistical details).

Lethal phase and anatomical analysis of fru lethal mutants:
Viability and lethal phase determination: To determine the lethal phase for various fru mutant genotypes, pupae and larvae were collected from the appropriate cross, staged, and allowed to develop. For genotypes where animals reached pharate adult stages, the operculum was removed from pupae and they were allowed to emerge, or they were dissected out of the pupal case. The adults obtained were kept in small groups on agar plates supplemented with yeast. In some cases, their behavior was videotaped.

General anatomical and neuroanatomical analysis: fru mutant animals were subjected to several types of anatomical analysis to characterize morphological phenotypes. For external cuticle preparations, adult or late stage pharate adult animals were fixed in alcohol, macerated in hot 10% KOH, dehydrated through an alcohol series and xylene, and mounted between two coverslips in Permount (SZABAD 1978 ). To examine the motorneuronal innervation of abdominal and genital muscles, adult abdomens were dissected, fixed in 4% paraformaldehyde in PBS, and prepared for immunohistochemistry. Nerve terminals were labeled using an anti-synaptotagmin antibody (1:1000 in PBS, 0.1% Triton-X and 2–10% normal goat serum; LITTLETON et al. 1993 ) to label synapses by the protocol described in FINLEY et al. 1997 . Since all fru lethal genotypes survived past the third instar larval stage, we examined the division pattern of a set of sex-specific abdominal neuroblasts by 5' bromodeoxyuridine (BrDU; Sigma, St. Louis) incorporation. Dissected larval and early pupal CNSs were incubated in BrDU for 4 hr and then processed for immunohistochemistry according to techniques in TAYLOR and TRUMAN 1992 .

All immunohistochemical analyses used biotinylated secondary antibodies and ABC reagents (Vector kit; Vector Laboratories, Burlingame, CA). The color reaction used diaminobenzidine for visualization according to standard techniques (TAYLOR and TRUMAN 1992 ; TAYLOR and KNITTEL 1995 ).

Muscle of Lawrence analysis of fru mutants: Procedures used for the characterization of the muscle of Lawrence (MOL) are detailed elsewhere (GAILEY et al. 1991 ). In brief, dorsal abdominal cuticles were dissected out, fixed, and cleared in methyl salicylate. Musculature was then visualized by birefringence in polarized light. Individual MOL phenotypes were established by a ranking system outlined in VILLELLA et al. 1997 ; see also Table 8 legend.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetic and molecular analysis of the new fru mutants
Isolation of new fru mutants: To generate new, more severe fru mutations, we mutagenized the fru² and fru⁴ mutations (GAILEY and HALL 1989 ; CASTRILLON et al. 1993 ). Both of these fru mutants are due to P-element inserts: the fru² P element is located between the 5' exons of the P3 and P4 promoters and fru⁴ is between the 5' exons of the P2 and P3 promoters (Fig 1B; ITO et al. 1996 ; RYNER et al. 1996 ; GOODWIN et al. 2000 ). The fru² P element was mutagenized by X rays and the fru⁴ P element was mobilized using standard techniques to generate imprecise excisions (see MATERIALS AND METHODS). From these mutageneses, we recovered eight fru mutations.

Cytological and molecular characterization of new fru mutants: Cytological examination of salivary chromosomes showed that six of the eight new fru^- mutations had visible aberrations (Table 1). The Df(3R)fru^w10, Df(3R)fru^w24, and Df(3R)fru^w25 alleles are deficiencies uncovering the 91B region of the third chromosome. In(3LR)fru^w12 is an inversion and T(3;Y)fru^w9 is a translocation, each with a chromosomal break at 91B, the location of fru. T(3R;het)fru^w12 is an inversion-cum-translocation also broken in 91B. Putting the findings that three of the fru²-derived mutations are inversions or translocations with one breakpoint at the cytological location of fru together with the fact that these rearrangements arose simultaneously with the inactivation of the wild-type white gene in the fru² P element suggests that these three rearrangements are broken in the fru² P element.

Southern analysis was used to examine whether these three rearrangements were broken in the fru² P element. These experiments showed that In(3LR)fru^w12 and In(3R)fru^w27 had lesions within a 4-kb HindIII fragment at the 5' end of the white⁺ gene in the P element. T(3;het)fru^w9 has a lesion located within a 3-kb BamHI fragment, partially overlapping the proximal end of the 4-kb HindIII fragment. These data are thus consistent with the idea that the fru breakpoints associated with these three rearrangements are within the fru² P element. Thus these three rearrangements should separate the P1-3 fru promoters, but not the P4 promoter, from the common fru coding sequences.

We also molecularly mapped the tiny deletions associated with the fru mutations Df(3R)fru^4-40 and Df(3R)fru^sat15. Df(3R)fru^4-40 was found to extend distally from within the fru⁴ P element for at least 70 kb (see MATERIALS AND METHODS). Df(3R)fru^sat15, which is derived from the excision of a P element inserted into the fru gene (ITO et al. 1996 ), extends centromere-proximal from within the fru^sat insert and removes the entire common fru coding region as well as several adjacent genes (Fig 1; see MATERIALS AND METHODS).

Transcript analysis of fru mutants: Since the new fru mutants described above are either deletions of portions of the fru locus or inversions or translocations that separate some fru promoters from fru coding sequences, these mutants are expected to lack various subsets of fru transcripts. They can thus be used to dissect the functions of various fru transcript classes. Rather than rely on results of the above cytological and molecular characterizations to infer which transcripts are produced by these mutant chromosomes, we used reverse transcription (RT)-PCR to directly examine the arrays of transcripts produced by various genotypes [In(3R)fru^w27/Df(3R)fru^w24, T(3R;het)fru^w12/ Df(3R)fru^w24, T(3R;het)fru^w12/ In(3R)fru^w27, Df(3R)fru^sat15/ Df(3R)fru^4-40], and Df(3R)Cha^M5/ T(3R;het)fru^w12 mutant animals and Df(3R)fru^w24/+ adults; the latter were used as a positive control]. RNA was extracted from these genotypes and reverse transcribed. Following the RT reactions aliquots of each reaction were amplified with various primer pairs (see MATERIALS AND METHODS) to determine which transcript classes were produced in each genotype.

We first determined whether any fru transcripts were produced in these genotypes by using primers that amplified a sequence from the protein-coding region common to all known transcript classes. This experiment detected fru transcripts in all five mutant genotypes (Fig 2A). These findings are as expected, since the region between P4 and the common coding region is normal on one of the chromosomes in each of these genotypes.

View larger version (64K): In this window In a new window Download PPT slide   Figure 2. RT-PCR analysis of fru mutants for the detection of sex-specific and sex-nonspecific transcripts. In A–G, lanes are as follows: lane 1, fruw27/fruw24; lane 2, fruw12/fruw24; lane 3, fru12/fruw27; lane 4, fruw24/+; lane 5, fur4-40/frusat15. Arrows point to the fru-specific bands. F, female-specific product; M, male-specific product. (A) The 317-nt RT-PCR product amplified from the fru common protein-coding region, which amplifies transcripts from all of the fru promoters (primers fru-25 and fru-26-rev). The slightly larger product detected is due to contamination of genomic DNA, which contains a 72-nt intron. Transcripts containing fru-coding sequences were detected in all fruw24/fruw27, fruw12/fruw24, fruw12/fruw27, and fru4-40/frusat15 mutants. Midstage pupae were used for fruw24/fruw27, fruw12/fruw24, and fruw12/fruw27 and adults were collected for the control. (B) The 305-nt RT-PCR product of female-specific P1 transcripts (primers fru-2 and fru-7-rev). Transcripts were detected in control fruw24/+ but no P1 transcripts were detected in any of the mutants. (C) The 1931- and 341-nt RT-PCR products amplified from P1 transcripts using primers (fru-31 and fru-26-rev) to amplify a sequence, which includes the male-specific splice to the common coding region. In the control fruw24/+ lane, both male-specific (lower band) and female-specific transcripts (upper band) are detected with these primers. No wild-type P1 transcripts were detected in any of the mutants. Bands in the fruw27/fruw24, fruw12/fruw27, and fru4-40/frusat15 lanes are not the expected size and are likely to be artifacts of the PCR. (D) The 379-nt RT-PCR product amplified from P2 transcripts (primers fru-29 and fru-55-rev-Xho). P2 transcripts were only detected in the control fruw24/+ lane. A faint band was present in the frusat15/fru4-40 lane but is smaller than the expected size and so is likely an artifact. (E) The 255-nt RT-PCR product amplified from P3 transcripts (primers fru-12 and fru-7-rev). P3 transcripts were not detected in fruw12/fruw24 animals. (F) The 360-nt RT-PCR product amplified from P4 transcripts (primers fru-40 and fru-26-rev). The fruw24/fruw27, fruw12/fruw24, fruw12/fruw27, and fru4-40/frusat15 mutants make P4 fru transcripts. (G) The 438-nt male P1 and 304-nt female P1 RT-PCR products amplified from total RNA of ChaM5/W12 adult flies (male primers fru-28 and fru-26-rev, female primers fru-2 and fru-7-rev). An 230-nt product is also detected with the female primers and may represent use of a 5' splice site 73 nt upstream of the female-specific 5' splice site. Note this product is detected in both males and females.

Whether the sex-specifically spliced P1 transcripts were produced in these mutants was tested with two sets of primers. One pair amplified the female product (Fig 2B) and another pair amplified both the male and female-specific products (Fig 2C), since the sequence of the P1 transcripts in males overlaps that in females. These primers amplified the expected products from control animals (Fig 2B and Fig C). No P1 transcripts were detected from In(3R)fru^w27/ Df(3R)fru^w24, Df(3R)fru^w24/ T(3R;het)fru^w12, T(3R;het)fru^w12/ In(3R)fru^w27, and Df(3R)fru^4-40/ Df(3R)fru^sat15 animals (Fig 2B and Fig C; data not shown). These findings are consistent with expectations since In(3R)fru^w27 and T(3R;het)fru^w12 are broken between P1 and the common coding sequences, Df(3R)fru^sat15 deletes the common coding sequences, Df(3R)fru^4-40 deletes the P1 promoter, and Df(3R)fru^w24 deletes the entire fru locus. The absence of P1 transcripts in these mutant combinations was confirmed by carrying out two rounds of nested PCR with primers that amplified the female product (data not shown).

Surprisingly, P1 transcripts were detected in the Df(3R)Cha^M5/T(3R;het)fru^w12 mutant combination (Fig 2G) with both sets of primers. As shown in other mutant combinations, T(3R;het)fru^w12 does not produce P1 transcripts spliced to the common coding region and thus these transcripts must come from the Df(3R)Cha^M5 chromosome. Df(3R)Cha^M5 is a large deletion beginning between the P1 and P2 promoters and extending distally far beyond the P1 promoter. Thus the RT-PCR product detected cannot be a fru transcript initiated from the P1 promoter (Fig 1B). However, the end of the Df(3R)Cha^M5 deletion that is within fru is located in the same large genomic restriction fragment that contains the 5' primer sequences used for the RT-PCR detection of P1 transcripts (Fig 1B). The results showing that RT-PCR products were detected with both the male and female primers indicate that the sequences corresponding to these primers are not deleted. Note that the authenticity of all PCR products was confirmed by Southern analysis using probes that did not contain the primer sequences (see MATERIALS AND METHODS). Therefore the most likely explanation for the "P1" transcripts detected in Df(3R)Cha^M5/T(3;het)fru^w12 is that these transcripts are being produced by an ectopic promoter that was juxtaposed to fru by the Df(3R)Cha^M5 deletion. If this is the case, these P1 transcripts may not be being expressed in the cells in which the P1 fru promoter is normally active. Indeed, antibodies specific to the fru male-specific proteins expressed from the P1 promoter fail to detect these proteins in whole mounts of central nervous systems of Df(3R)Cha^M5/Df(3R)P14 at the pupal stages when these proteins are expressed in wild type (LEE et al. 2000 ).

In addition to the sex-specific P1 products detected in Df(3R)Cha^M5/T(3R;het)fru^w12, a sex-nonspecific product, slightly smaller than the wild-type product, is detected in both sexes with the female-specific primers. There is a 5' splice site consensus sequence upstream of the female 5' splice site that would give a product of this size if it were used. Sex-nonspecific products using this splice site have not been detected in wild type or in any of the other mutants and thus are specific to the Df(3R)Cha^M5 deficiency. There are two possible explanations for why this site might be used in Df(3R)Cha^M5-derived transcripts. Because this shorter product is produced in both sexes, it may be the result of ectopic promoter-driven expression in tissues where this cryptic splice site can be utilized. Alternatively, the Df(3R)Cha^M5 deletion may remove upstream sequences that affect the use of this splice site. The latter explanation seems less likely because the primary P1 transcript from Df(3R)Cha^M5 contains sequences (the male-specific PCR primer) upstream of the male 5' splice site, which is >1.6 kb away from the regulatory sequences that are known to affect the regulation of the female-specific splice site in wild type (HEINRICHS et al. 1998 ). These results indicate that a novel fru transcript is produced from the Df(3R)Cha^M5 deficiency chromosome; whether this transcript is expressed in the proper cells to supply fru function is unknown, but phenotypically there is no evidence for P1 function (see below).

On the basis of the location and nature of the fru rearrangements in these genotypes, transcripts initiated at the P2 promoter and spliced to the common coding region are not expected to be present except in the case of the Df(3R)Cha^M5/ T(3R;het)fru^w12 genotype. As predicted, no P2 transcripts were detected in any of the fru mutant genotypes tested except for Df(3R)Cha^M5/ T(3R;het)fru^w12 (Fig 2D, data not shown). In the latter genotype the Df(3R)Cha^M5 allele is the likely source of the P2 RT-PCR products since it does not delete the P2 promoter (Fig 1B); and the T(3;het)fru^w12 allele alone, as shown in the Df(3R)fru^w24/T(3R;het)fru^w12 lane (Fig 2D), does not produce a P2 product spliced to the common coding region.

With respect to the P3 promoter, the locations of the breakpoints in these fru alleles predict that only the Df(3R)Cha^M5 and Df(3R)fru^4-40 chromosomes should produce P3 transcripts spliced to the common coding region. These expectations are fulfilled in all but one case (Fig 2E). Unexpectedly, in all genotypes involving the In(3R)fru^w27 allele, RT-PCR revealed P3 transcripts. That these products are real is indicated by the facts that they are of the correct size and that they contain the expected fru mRNA sequences between the PCR primer pairs (see MATERIALS AND METHODS). These findings suggest that In(3R)fru^w27 is more complex than indicated by its cytological and molecular characterization. The In(3R)fru^w27 chromosome has one inversion with a break in fru and a second inversion that does not involve fru. Southern analysis indicated that there was a lesion within the white gene of the fru²-associated P element on which In(3R)fru^w27 was induced, and we had inferred that this was the breakpoint of the In(3R)fru^w27-associated inversion. However, the RT-PCR findings that P3 but not P2 transcripts are produced from the In(3R)fru^w27 chromosome suggest that there is a breakpoint in the region between the P2 and P3 promoters as shown in Fig 1B.

In the case of the P4 promoter, all the fru alleles analyzed, except Df(3R)fru^sat15 and Df(3R)fru^w24 (both of which delete P4), are expected to produce P4-derived transcripts spliced to the common coding region. The results (Fig 2F) are in accord with these expectations.

In summary, the above data show that different fru mutants are missing different subsets of fru transcripts (Table 2). As described above, our results are consistent with the idea that Df(3R)Cha^M5 affects only P1 function. Two mutations, Df(3R)fru^4-40 and In(3R)fru^w27, lack both P1- and P2-derived transcripts. One mutation, T(3;het)fru^w12, lacks P1-, P2-, and P3-derived transcripts, and we infer that T(3;Y)fru^w9 does also on the basis of the location of its breakpoint (see above). Finally, Df(3R)fru^sat15, Df(3R)fru^w24, and the previously characterized deletions Df(3R)P14 and Df(3R)Cha^M7 lack functional products from all fru promoters but also delete neighboring genes. By using different combinations of these fru alleles, we have begun to associate specific fru mutant phenotypes with the loss of different fru transcript classes.

Phenotypic analysis of new fru mutants
Origin of fru's vital function: To determine which fru promoters encode its vital function(s) we examined the viability of various heteroallelic fru mutant combinations that lacked particular subsets of fru transcripts (Table 3).

The finding that Df(3R)Cha^M5/ Df(3R)fru^sat15 individuals, which have little or no P1 function and express the products of the other fru promoters, have normal viability (Table 3B) indicates that the products of the P1 fru promoter are not needed for viability. This conclusion is also supported by the findings that Df(3R)Cha^M5 is also viable over Df(3R)P14 (GAILEY et al. 1991 ). Df(3R)P14 is like Df(3R)fru^sat15 in that it is a deletion that has one end in the middle of fru and extends proximally, deleting all fru coding sequences as well as neighboring genes.

The RT-PCR experiments described above suggest there are two aberrations, Df(3R)fru^4-40 and In(3R)fru^w27, which lack P1 and P2 function, but have P3 and P4 function. However, these two aberrations give completely discordant results when used in complementation tests with other fru alleles that lack fru's vital function. Thus Df(3R)fru^4-40 over Df(3R)fru^sat15 or Df(3R)P14 is fully viable (Table 3C), whereas In(3R)fru^w27 over Df(3R)fru^sat15, Df(3R)Cha^M7, Df(3R)fru^w24, T(3;het)fru^w12, or T(3;Y)fru^w9 is lethal (Table 3D). As explained in the DISCUSSION we believe that this disparity is due to In(3R)fru^w27 being defective in more than just fru's P1 and P2 functions and that the Df(3R)fru^4-40 results are reflective of the phenotype resulting from the loss of both P1 and P2 functions. This would indicate that P2-derived products, like those derived from P1, are not needed for viability.

With respect to P3, our analysis suggests that both T(3;het)fru^w12 and T(3;Y)fru^w9 do not produce P3-, P1-, or P2-derived products. Both T(3;het)fru^w12 and T(3;Y)fru^w9 are lethal over the null fru mutants Df(3R)fru^sat15, Df(3R)fru^w24, and Df(3R)Cha^M7, as well as each other (Table 3D). These results indicate that the products of P3 are essential for viability.

We do not have a genotype with which we can assess the viability effects of losing P4 but not P3 function. While we examined the viability effects of genotypes that are complete nulls for fru (Table 3E), and these are invariably lethal as expected, these deficiency combinations are also deleted for genes flanking fru. Nevertheless the phenotypic characterization of the lethal phenotypes in these deficiency combinations (see below) is useful as it places an upper boundary on the severity of the phenotype that can be expected from a null fru genotype.

Male sterility of the fru mutants: Previous studies established that males carrying many fru mutants and mutant combinations are sterile and that this sterility is behavioral in origin: Such males frequently fail to copulate (GAILEY and HALL 1989 ; GAILEY et al. 1991 ; VILLELLA et al. 1997 ). In addition, it was recently found by examining fru mutant combinations that do sometimes mate that fru function is also required during mating for successful insemination (LEE et al. 2001 ). To gain insight into which fru products are required for male fertility we assayed the effects of the new fru alleles on fertility.

That T(3;Y)fru^w9, T(3;het)fru^w12, Df(3R)fru^w24, In(3R)fru^w27, Df(3R)fru^sat15, and Df(3R)fru^4-40 all abolish male fertility is shown by the complete sterility of all combinations of these mutants tested with each other (Table 4D). Since all the genotypes in Table 4D, lack the functions encoded by the P1 and P2 fru promoters, these results indicate that some product(s) from these promoters are required for male fertility. To examine whether the products of the P1 promoter were required for male fertility we examined the effects of trans-heterozygotes between Df(3R)Cha^M5 and T(3;Y fru^w9, T(3;het)fru^w12, In(3R)fru^w27, Df(3R)fru^4-40, or Df(3R)fru^sat15. With one exception, all combinations were completely sterile (Table 4C). The exception was that in one round of tests 3/65 Df(3R)Cha^M5/In(3R)fru^w27 males were fertile, whereas several subsequent rounds of tests found 0/36 fertile. With the exception of these 3 males all the results in Table 4C and Table 4D, are consistent with the hypothesis that functional products of the P1 fru promoter are essential for male fertility.

We also examined the effects on male fertility of the new fru mutations in combination with the hypomorphic alleles fru¹, fru², fru³, and fru⁴ (Table 4E). The fru², fru³, and fru⁴ alleles are due to P-element insertions in fru and have reduced levels of either P1- and P2-derived transcripts (fru³ and fru⁴) or reduced levels of P1-, P2-, and P3-derived transcripts (fru²). P4 transcripts are not affected by these mutations (GOODWIN et al. 2000 ). fru¹ is an inversion that removes sequences upstream of the P1 promoter and results in an altered pattern of expression of P1 in the CNS (GOODWIN et al. 2000 ). The fru³ and fru⁴ alleles show more severe behavioral defects than fru¹ or fru² (VILLELLA et al. 1997 ). Consistent with the latter observation T(3;Y)fru^w9, T(3;het)fru^w12, Df(3R)fru^w24, In(3R)fru^w27, Df(3R)fru^4-40, and Df(3R)fru^sat15 are all completely sterile over either fru³ or fru⁴, whereas a few combinations of these alleles with either fru¹ or fru² show some male fertility. Two of the latter genotypes show results worthy of note: Most fru¹/Df(3R)fru^sat15 and fru²/Df(3R)fru^sat15 males are fertile. This is a surprising result, since males with either fru¹ or fru² over other null deletions of fru [e.g., Df(3R)fru^w24, Df(3R)Cha^M7, and Df(3R)P14] were invariably sterile (GAILEY and HALL 1989 ; VILLELLA et al. 1997 ). This may reflect genetic background effects.

As controls for the above experiments we examined the fertility of the new fru alleles when heterozygous with a wild-type chromosome and two deficiencies flanking fru (Table 4A). Two flanking deletions, gl^BX7, a glass allele associated with a 20-kb deletion centromere-proximal to fru, and Df(3R)148.5-1, a large deletion centromere-distal to fru, complemented the sterility phenotype of these new fru lesions, showing that there are no additional male-sterile mutations in the vicinity of fru on these chromosomes.

Male-male and male-female courtship performed by fru mutant males: To gain more insight into fru's male-specific function, we examined various components of male courtship behavior in a number of heteroallelic fru mutant combinations. One measure of overall courtship levels is given by the courtship index (CI), which indicates the amount of time a male fly spends courting (see MATERIALS AND METHODS). The wing extension index (WEI) provides a second measure of the courtship (see MATERIALS AND METHODS). Wild-type males court females vigorously, but court males at very low levels, as is seen from the CIs when T(3;Y)fru^w9, T(3;het)fru^w12, Df(3R)fru^w24, In(3R)fru^w27, or Df(3R)fru^4-40 are heterozygous with a wild-type chromosome (Table 5A). During courtship wild-type males perform courtship song for roughly one-half of the interval spent courting (WEI; Table 5A).

Previous analyses (RYNER et al. 1996 ; GOODWIN et al. 2000 ) had led us to suggest that it was the products of the P1 fru transcripts that carried out fru's role in male courtship. To assess the effects on courtship of just-impaired P1 function we examined the effects of T(3;Y)fru^w9, T(3;het)fru^w12, Df(3R)fru^sat15, and In(3R)fru^w27 when heterozygous with Df(3R)Cha^M5 (Table 5B). We also examined the effects of the absence of both P1- and P2-encoded functions on male courtship behavior in T(3;Y)fru^w9, T(3;het)fru^w12, In(3R)fru^w27, Df(3R)fru^sat15, and Df(3R)P14 when heterozygous with Df(3R)fru^4-40 (Table 5C). In all of these genotypes male-female and male-male courtship measured by either the CI or the WEI is almost completely abolished [unlike what is seen in less severe fru genotypes (see below)]. Taken together these results indicate that in the absence of fru's P1-encoded functions all aspects of male-male and male-female courtship are abolished.

We also examined male courtship behavior in flies in which the new fru alleles were heterozygous with one of the hypomorphic alleles, fru¹, fru², fru³, or fru⁴ (Table 5D). The new alleles over either fru³ or fru⁴ were nearly as severe in their effects as the above genotypes in which fru's P1 function was absent: Male-female courtship as measured by the CI or WEI was abolished as was male-male courtship as measured by WEI. However, in some genotypes, most notably Df(3R)fru^4-40/fru³ and Df(3R)fru^4-40/fru⁴, there was significant male-male courtship as measured by the CI. The effects of the new alleles over fru² were also severe: male-male courtship was essentially abolished as measured by either the CI or WEI, as was male-female courtship as measured by the WEI. In most genotypes involving fru², measurable male-female courtship was produced. Courtship behavior of males carrying one of the new fru alleles over fru¹ with either a male or female was substantial as measured by both CI and WEI. These males had CIs similar to those of fru¹ homozygote males (VILLELLA et al. 1997 ). Interestingly, these fru¹ heterozygous combinations are almost all sterile (Table 4B). In all fru¹ heterozygous combinations male-female courtship was less frequent than that seen with wild-type males (Table 5D vs. 5A); this may be partially due to these mutant individuals being in general less active than wild type (see below).

In summary, heteroallelic combinations of all of the new fru mutant alleles with either fru², fru³, fru⁴, Df(3R)fru^4-40, or Df(3R)Cha^M5 were not significantly different from wild-type males in their courtship of other males, showing that they no longer had the strong male-male courtship behavior associated with fru¹ or homozygosity for the P-element fru alleles (Table 5; VILLELLA et al. 1997 ). Thus these fru males are significantly different from wild-type males in that they do not court females and different from fru¹, fru³, and fru⁴ homozygotes since they do not court males and females. These findings also show that the reduction in wing extension first observed in fru³ and fru⁴ homozygotes and trans-heterozygotes (VILLELLA et al. 1997 ) is common to other severely affected fru genotypes as well.

Although most fru males that court respond to males and females roughly equally, several previously studied fru mutant genotypes showed a courtship bias toward either males or females. For example, fru³ males courted males more vigorously than females while fru² males courted females more avidly than males (GAILEY and HALL 1989 ; VILLELLA et al. 1997 ). We therefore asked whether these new fru mutants showed any consistent bias in the courtship of males vs. females. Males from twenty-four fru mutant genotypes courted males and females approximately equally (Table 5). Overall, the CIs of males of a particular fru genotype to male or female targets are highly correlated [CI_m-m compared to CI_m-f: F_{(29, 4586)} = 64.9, correlation coefficient = 0.83, R² = 69.1%; linear regression analysis, Statgraphics 5.0], suggesting that fru males are not discriminating between male and female partners independent of whatever level of courtship they generate.

Short-term activity of fru mutant males: One explanation for the failure of fru males to court is that these males might be generally inactive or sluggish (for review, see HALL 1994 ). To evaluate the general behavioral robustness of the new fru genotypes, a short-term activity assay was used to measure the voluntary activity a single male performs in the same chamber used for the courtship assays (VILLELLA et al. 1997 ). Tests of five genotypes that do not express P1 or P2 fru transcripts [Df(3R)fru^4-40 heterozygous with either T(3;Y)fru^w9, T(3;het)fru^w12, Df(3R)fru^sat15, In(3R)fru^w27, or Df(3R)P14] and four genotypes that have little or no functional P1 products [Df(3R)Cha^M5 heterozygous with either T(3;Y)fru^w9, T(3;het)fru^w12, Df(3R)fru^w27, or Df(3R)fru^sat15] showed that nearly all of these genotypes had reduced short-term activity compared to fru⁺ control males (Table 6B vs. A). However, two of these genotypes [Df(3R)fru^4-40/Df(3R)P14 and Df(3R)Cha^M5/In(3R)fru^w27] had wild-type activity levels, suggesting that the reduced activity seen in the other genotypes may be due to background effects rather than the result of reduced fru expression.

We also examined short-term activity in males heteroallelic for one of the new fru alleles and either fru¹, fru², fru³, or fru⁴. Heteroallelic combinations with fru³ or fru⁴ had, with one exception, wild-type levels of activity (Table 6C). However, all heteroallelic combinations of one of the new alleles with fru² showed greatly reduced short-term activity (Table 6C). Since all of the fru^w- alleles tested were induced on a fru² parental chromosome, the reduced activity seen in these genotypes likely represents the effects of this common genetic background. In this regard it is worth noting that In(3R)fru^w27/(T(3;het)fru^w12 escapers (this is an essentially lethal genotype; Table 3) also show low levels of activity.

Overall, almost one-half of the fru heteroallelic combinations tested were significantly less active than wild-type males (Table 6B and Table 6C). However, there is only a weak correlation between the level of short-term activity for these fru mutant males and the mean CIs they produce for either male-male or male-female courtship (Table 6 legend). These findings suggest that the failure of fru mutant males of these genotypes to court prospective mates in pairwise tests is predominantly a specific courtship defect rather than simply the result of these males being less active.

Male-male group courtship (chaining) by fru mutant males: fru mutant males exhibit a dramatic behavioral phenotype, the formation of male-male courtship chains (HALL 1978 ; GAILEY and HALL 1989 ; GAILEY et al. 1991 ; VILLELLA et al. 1997 ). Males homozygous for fru¹ court each other in groups, from a young-adult age onward (VILLELLA et al. 1997 ). For other genotypes, such as fru³ and fru⁴, significant chaining is seen only after the males have been grouped together for a few days (VILLELLA et al. 1997 ). A ChI, measuring the time that three or more males were courting during the observation period, was calculated to quantify the amount of intermale courtship produced by the different genotypes we examined (Table 5).

Males carrying one of the new fru alleles [T(3;Y)fru^w9, T(3;het)fru^w12, Df(3R)fru^w24, In(3R)fru^w27, and Df(3R)fru^4-40] over one of the hypomorphic fru alleles (fru¹, fru², fru³, and fru⁴) exhibited substantial chaining (Table 5D). Chaining was highest in fru¹ heteroallelic combinations, somewhat less in fru³ and fru⁴ heteroallelic combinations, and lower still in fru² heteroallelic combinations. This parallels the levels of chaining seen in homozygotes for these four hypomorphic alleles (VILLELLA et al. 1997 ). While the levels of chaining in fru¹ heteroallelic combinations are lower than that seen in fru¹ homozygotes [41–58% (Table 5D) vs. 70% (VILLELLA et al. 1997 )], in the cases of the fru², fru³, and fru⁴ heteroallelic combinations the levels of chaining are, in general, comparable to those previously reported for the respective homozygous mutants (VILLELLA et al. 1997 ).

Chaining was also examined in four heteroallelic fru combinations that have little or no functional P1 fru promoter-derived products (Table 5B) and six heteroallelic combinations that have no functional P1 or P2 fru promoter-derived products (Table 5C). The results were quite variable with four genotypes [T(3;het)fru^w12/Df(3R)Cha^M5, T(3;het)fru^w12/Df(3R)fru^4-40, T(3;het)fru^w12/In(3R)fru^w27, and Df(3R)fru^4-40/Df(3R)P14], showing no chaining, while the remaining five genotypes [T(3;Y)fru^w9/Df(3R)Cha^M5, In(3R)fru^w27/Df(3R)Cha^M5, T(3;Y)fru^w9/Df(3R)fru^4-40, In(3R)fru^w27/Df(3R)fru^4-40, and Df(3R)fru^4-40/Df(3R)fru^sat15] showed significant levels of chaining. We do not understand the bases for these differences between ostensibly equivalent fru genotypes. Three of the four genotypes that did not chain are less active than wild type (Table 6), but one of the fru mutants that did not chain, Df(3R)fru^4-40/Df(3R)P14, had high levels of short-term activity. On the other hand, some mutant genotypes with low mean short-term activity [In(3R)fru^w27/Df(3R)fru^4-40 and T(3;Y)fru^w9/fru²] still performed moderate levels of chaining (Table 5 and Table 6).

To determine whether there was a general relationship between the level of short-term activity and chaining behavior by fru mutant males, the correlation between the mean ChI vs. the mean short-term activity measurement was calculated for all genotypes (correlation coefficient = 0.41, R² = 17.2%; linear regression analysis, Statgraphics 5.0). It is apparent from the regression analysis that there is a correlation between the level of short-term activity of flies and their ChI. Nevertheless, these findings do suggest that a part of the reduction in chaining levels represents the loss of a fru-dependent function and that the reduction in ChI is not solely due to the reduced activity of the participating male flies.

In the analysis of the courtship behavior of these fru mutant males, we noted that males of several fru genotypes that had essentially no courtship to either males or females in pairwise tests still showed male-male chaining behavior at moderate to high levels (Table 5). To better analyze the relationship between the different measures of courtship activity, a linear regression analysis was carried out. It shows that there is a strong positive correlation between the mean CI_m-m and the ChI [F_{(29, 3590)} = 34.18, correlation coefficient = 0.74; R² = 54.1%; linear regression analysis, SAS). Seven fru mutant genotypes [e.g., Df(3R)fru^4-40/Df(3R)P14, T(3;het)fru^w12/Df(3R)fru^4-40, T(3;het)fru^w12/Df(3R)Cha^M5], which showed no chaining behavior, also did not court males or females, suggesting that the loss of the chaining behavior occurred separately and subsequently to the loss of all of the individual courtship behaviors as measured in courtship tests and that the absence of chaining is an indication of a more severe loss of fru function.

Expression of the courtship song by fru mutant males: Wild-type males produce a dual component song consisting of a hum or sine song and a pulse song composed of a train of pulses, which is amenable to detailed quantitative analysis (WHEELER et al. 1988 , WHEELER et al. 1989 ; BERNSTEIN et al. 1992 ). Control fru^w-/+ males generated normal courtship song except that the width of the fast-Fourier transform (FFT) was smaller in the control males vs. wild type (Table 7A). fru^1-4/+ controls also showed a similar change in the width of the FFT compared to wild-type males (VILLELLA et al. 1997 ).

Males carrying one of the new fru alleles over fru¹ produced a courtship song that had longer interpulse intervals (IPIs), but essentially normal intrapulse frequencies, cycles per pulse, and width of the FFT when compared to the control males (Table 7B).

Previous experiments had shown that males mutant for some fru alleles, such as fru³/fru¹, produced a courtship song, but homozygous fru³ males or fru³/P14 males did not (RYNER et al. 1996 ; VILLELLA et al. 1997 ). We examined the song production of males carrying heteroallelic combinations of the new fru alleles with either fru³ or fru⁴ as a further test for the severity of the new fru alleles. These males did not produce a courtship song even though they did produce a small amount of wing extension (Table 5D and Table 7B). Other heteroallelic combinations with either Df(3R)Cha^M5 or Df(3R)fru^4-40, which had very reduced wing extension, also failed to produce courtship song during the times that they did extend their wing (Table 7C). These findings indicate that the new mutants are stronger alleles of fru and that courtship song is dependent on P1 function.

MOL of fru mutant males: fru was previously shown to control the male-specific differentiation of the MOL (GAILEY et al. 1991 ). Five of the new fru mutations were tested in combination with previously reported fru mutations or deletions for their effects on MOL formation (Table 8). All viable combinations of T(3;Y)fru^w9, T(3;het)fru^w12, Df(3R)fru^w24, and In(3R)fru^w27 with fru³, fru⁴, Df(3R)Cha^M5, and Df(3R)fru^4-40 failed to make a normal-sized MOL; all of the muscle fibers were small and the same size as nearby longitudinal muscles. This represents one of the strongest fru-associated MOL phenotypes (GAILEY et al. 1991 ; TAYLOR and KNITTEL 1995 ; VILLELLA et al. 1997 ; B. J. TAYLOR and L. KNITTEL, unpublished results).

Trans-heterozygotes between Df(3R)Cha^M5 and T(3;Y)fru^w9, T(3;het)fru^w12, or In(3R)fru^w27 showed almost no MOL development (Table 8). The slight MOL development noted in these genotypes likely stems from the ad hoc distinction of an occasional MOL showing some coalescence of the small longitudinal fibers at the expected fifth-abdominal-tergite site of MOL organization, the "a-1" category of VILLELLA et al. 1997 . It is not clear that these infrequent observations of marginal fiber clustering correlate with fru gene expression. Thus these results indicate that P1-derived functions are also essential for MOL development.

T(3;Y)fru⁹, T(3;het)fru¹², or In(3R)fru^w27 in combination with the hypomorphic alleles fru¹, fru², fru³, or fru⁴ had varying effects on MOL development. The fru³ and fru⁴ combinations showed the most severe disruption and were indistinguishable from the Df(3R)Cha^M5 combinations (MOLs missing and sterile; Table 8 and Table 4B). However, the fru¹ and fru² combinations showed an interesting difference. Whereas these combinations also led to sterility (except for the marginal case of fru²/T(3;Y)fru^w9; Table 4E), they resulted in an appreciable amount of intermediate-to-complete MOL development (Table 8). These results confirm that among all fru alleles, the fru¹ and fru² mutations have the weakest impact on the MOL, with fru¹ homozygotes showing a variably intermediate phenotype that is not exacerbated when heterozygous with any other fru allele, while fru² homozygotes develop a nearly wild-type MOL (rows 1 and 2, Table 8; VILLELLA et al. 1997 ).

Characterization of fru's essential functions
Lethal phase: To define the development stages when fru's vital functions are needed and the biological roles of the P3 and P4 transcripts, we examined animals from seven fru mutant genotypes chosen to reflect the range of potential lethal phases. Animals of five fru genotypes, which lacked P1, P2, and P3 function, such as T(3;het)fru^w12/Df(3R)fru^sat15 and T(3;het)fru^w12/Df(3R)fru^w24, reached the adult stage but were generally unable to eclose from the pupal case (Table 9A, Table 3C, and Table 3D). On close inspection, both males and females that died within the pupal case had attempted to eclose. These animals were able to successfully emerge as adults if dissected from the pupal case (Table 9A). Such freed animals performed posteclosion behaviors, such as wing and proboscis expansion, but were unable to fully extend their wings, which projected from the body at a 70–90° angle (Fig 3A). Once they emerged, they routinely survived for 7–14 days. By a number of phenotypic criteria, T(3;het)fru^w12/Df animals were more severely affected than the other genotypes from which survivors were obtained: Fewer T(3;het)fru^w12/Df animals reached the adult stages, and they had more severe anatomical phenotypes (see below) and shorter survival times.

View larger version (141K): In this window In a new window Download PPT slide   Figure 3. Anatomical defects in fru mutant animals. (A) Photograph of a fruw12/ fruw27 female, which had been released from the pupal case, showing the partially expanded wings (arrow) and bent tibial segment of the metathoracic leg (lower arrow). (B) Photographs of a metathoracic leg (3rd) from a cuticle preparation of a fruw27/ fruw24 female with a defective tibial (tib) first tarsal joint (tar; arrow). (C) Photographs of legs from a cuticle preparation of a fru1/ fruw24 female with normal leg joints. Upper leg is a mesothoracic leg (2nd) and middle and lower legs are left and right metathoracic (3rd) legs. Bar for B and C, 100 µm. (D) Photograph of the ventral side of frusat15/fruw24 (left) and Canton-S (right) 2- to 4-hr prepupae. The frusat15/fruw24 animal has not retracted its head into the typical prepupal form and the anterior (upper arrow) and posterior spiracles (lower arrow) have not fully everted. The mutant prepupa is more elongated than the wild-type prepupa. These fru mutants typically die a few hours later after the formation of the posterior bubble.

Since there are currently no genotypes that are null for just fru, we examined the lethal phases of three genotypes [Df(3R)fru^sat15/Df(3R)fru^sat15, Df(3R)fru^sat15/Df(3R)P14, and Df(3R)fru^sat15/Df(3R)fru^w24] that are null for fru but that also delete a small number of adjacent genes as these genotypes provide an upper limit on the severity and nature of a fru null phenotype (Table 9B). These three fru genotypes died after forming prepupae (Fig 3D, Table 9B). Df(3R)fru^sat15 is a small deficiency that deletes all fru coding sequences as well as all or part of the six genes proximal to fru. Df(3R)fru^w24 and Df(3R)P14 are larger deficiencies that encompass Df(3R)fru^sat15 and so these three genotypes should be equivalent in terms of the genes they delete. Since Df(3R)fru^sat15 homozygotes have a slightly earlier lethal phase than either Df(3R)fru^sat15/Df(3R)P14 or Df(3R)fru^sat15/Df(3R)fru^w24, this must be due to other lesions on the Df(3R)fru^sat15 chromosome. In the most severely affected cases fru prepupae were tanned and frequently formed a posterior bubble but did not have everted anterior spiracles or retracted anterior segments, thus exhibiting some characteristics associated with wandering third instar larvae (BAINBRIDGE and BOWNES 1981 ; Fig 3D; Table 9B). To determine whether these mutants had a defined lethal phase or whether a fraction of the larvae simply died at each molt, we collected 51 first instar Df(3R)fru^sat15/Df(3R)fru^w24 larvae and monitored their progression through the other larval stages into early pupal development. All animals reached the prepupal stage, suggesting that this pupal lethal phase was specific. The early pupal lethality of these mutants is the most extreme phenotype that is potentially associated with the loss of the fru function. Two other genotypes, Df(3R)fru^4-40/Df(3R)fru^w24 and Df(3R)fru^w24/Df(3R)P14, deleted for larger regions encompassing fru, were examined and found to die at the second to third larval molt (data not shown; Table 3). It is likely that the larval lethal phase in these genotypes is due to the absence of other vital gene(s) located in the large regions flanking the fru locus uncovered by these deficiencies. Thus, the lethal phase associated with loss of fru function is during metamorphosis.

The adult escapers or released animals from these fru genotypes all had minor defects in their external appearance (Table 9C). In a high proportion of these mutant escapers, one or more imaginal discs failed to evert and after emergence it was common for one or both wings to not be fully expanded. When the wings were inflated, they were abnormally positioned, being held at a 70–90° angle from the body (Fig 3A; Table 9C). Furthermore, the femur was frequently bent and one or more joints were defective in at least one leg per animal, usually the femur/tibial or the tibial/first tarsal joint in the metathoracic leg (Fig 3B, Table 9C). These defective joints were enlarged and frequently had additional bristles. Based on cuticle preparations made of mutant flies, males and females had duplicated sternopleural and anterior scutellar macrochaetes (Table 9C). These anatomical defects were not found in the fully viable fru mutant combinations Df(3R)Cha^M5/In(3R)fru^w27 (male, n = 4; female, n = 3) and fru¹/Df(3R)fru^w24 (male, n = 5; female, n = 18; Fig 3C; data not shown).

Neuronal phenotypes of fru mutants: The inability of some of these new fru mutants to emerge from the pupal case led us to examine motorneuronal synapses on abdominal muscles to determine whether all muscles were innervated and whether their synaptic morphology was normal. Innervation was assessed by labeling abdominal carcasses with antibodies to the synapse-specific protein, synaptotagmin (LITTLETON et al. 1993 ). We found that all skeletal and visceral muscles in the abdomen of T(3;het)fru^w12/In(3R)fru^w27 (n = 11), Df(3R)fru^w24/In(3R)fru^w27 (n = 50), and fru^w12/fru^w24 (male, n = 2; female, n = 2) animals were innervated. However, fewer motorneuronal terminals were present on the lateral muscles of the abdomen. These muscles are arranged as a series of parallel muscle fibers from the lateral edge of the dorsal tergite to the ventral sternite (Fig 4). In wild-type males and females, these muscles are innervated by nerves that have extensive terminal arborizations onto several muscle fibers (Fig 4A and Fig B). In fru mutants that produce just some P3 and P4 or just P4 transcripts, there were fewer branches from the nerve terminal that arborized on these lateral abdominal muscles (Fig 4C and Fig D; H.-J. SONG and B. J. TAYLOR, unpublished results). In addition, in these fru mutants, the innervation of the lateral abdominal muscles was concentrated in the lateral and middle zones, near the dorsal side of the body, and did not penetrate more medial zones of these abdominal muscles as occurs in wild type. These differences in abdominal muscle innervation were not seen in fru¹ homozygote (n = 5) or Df(3R)Cha^M5/In(3R)fru^w27 trans-heterozygote (n = 6; Fig 4E) mutant flies, suggesting that this synaptic defect is a new fru non-sex-specific neuronal phenotype associated with lethal fru alleles.

View larger version (135K): In this window In a new window Download PPT slide   Figure 4. Photomicrograph of lateral A4 abdominal region labeled with anti-synaptotagmin to visualize motor nerve terminals. (A) Canton-S female showing that these muscles are innervated by motorneurons that make highly branched terminal arborizations (arrow). The longitudinal muscles along the ventral midline are indicated by Vm in this and subsequent figures. (B) Canton-S male showing that these muscles are innervated by motorneurons that make highly branched terminal arborizations. Bar for A and B, 50 µm. (C) A fruw12/ fruw27 female in which the abdominal motorneurons produce fewer lateral terminal arborizations (arrow) as they terminate on the lateral muscles. (D) A fruw27/ fruw24 male in which the abdominal motorneurons make many fewer terminal branches than wild type. (E) A ChaM5/fruw27 female with normal abdominal innervation. Bar for C–E, 50 µm.

The finding of these defective motorneuronal arborizations in these new fru^- mutants led us to ask whether there were any other obvious neuronal phenotypes. We examined postembryonic neurogenesis in Df(3R)fru^w24/In(3R)fru^w27 (n = 2), Df(3R)fru^w24/Df(3R)fru^sat15 (n = 3), and fru¹ (n = 4) third instar larvae by following the incorporation of BrDU into dividing neuroblasts (TRUMAN and BATE 1989 ; TAYLOR and TRUMAN 1992 ). No difference was detected in the number or pattern of divisions of larval neuroblasts in the thoracic ganglia or brain in fru mutants compared to wild-type males and females (data not shown). In addition, the male-specific abdominal neuroblasts (TRUMAN and BATE 1989 ; TAYLOR and TRUMAN 1992 ) were present and actively dividing in all of the fru mutants including Df(3R)fru^w24/Df(3R)fru^sat15 males (data not shown). Since Df(3R)fru^w24/Df(3R)fru^sat15 males and females die just after metamorphosis, the division of the neuroblasts in these mutants shows that the larval program for these neuroblasts is still functioning up until the time of the animals' death. No dividing abdominal neuroblasts were detectable in any fru mutant female CNSs from wandering third instar larvae, showing that there is no male transformation of the sex-specific abdominal neuroblasts in fru mutant females (data not shown). Although we cannot rule out more subtle defects, the overall division pattern of the postembryonic neuroblasts, which produces the bulk of adult-specific neurons, appears to be normal in fru mutants. Thus, of the phenotypes examined, the only neuronal defects detected in the fru mutant animals were reduced synaptic arborizations of motorneuronal terminals.

In situ expression pattern of fru transcripts in fru mutants: The sex-specific expression of fru in wild-type males and females is found almost exclusively in neuronal clusters in the brain and ventral nerve cord (RYNER et al. 1996 ; LEE et al. 2000 ). The designations of these clusters on the basis of in situ hybridization to fru transcripts in sections (RYNER et al. 1996 ) and staining of whole mounts with antibodies to FRU proteins or in situ hybridization to fru transcripts (LEE et al. 2000 ) differ. Here we give both designations. In the brain, males and females have a number of prominent clusters of labeled neurons in similar locations (LEE et al. 2000 ). We examined several of these for the presence of fru transcripts: a cluster distributed in the dorsal posterior protocerebrum, one cluster medial to the optic lobes, three groups in the dorsal anterior protocerebrum, and one near the antennal lobes (Fig 6; RYNER et al. 1996 ; GOODWIN et al. 2000 ; M. FOSS, G.-H. LEE, E. REYNAUD, L. C. RYNER, J. C. HALL, B. S. BAKER and B. J. TAYLOR, unpublished results). We also examined three male-specific groups of neurons in the thoracic-abdominal ganglion; one is located between the pro and mesothoracic neuromeres, the second is in the mesothoracic region, and the third is in the abdominal ganglion (Fig 6; RYNER et al. 1996 ; GOODWIN et al. 2000 ; M. FOSS, G.-H. LEE, E. REYNAUD, L. C. RYNER, J. C. HALL, B. S. BAKER and B. J. TAYLOR, unpublished results).

View larger version (84K): In this window In a new window Download PPT slide   Figure 5. In situ hybridization of antisense riboprobes to the P1 promoter (S box in Fig 1B) to sections of fruw24/fruw27 mutant male. (A) Low power view of a section through the head at the level of neuronal group 7 (LEE et al. 2000 ; group 6, RYNER et al. 1996 ; see Fig 6) near the antennal lobe. Bar, 60 µm; OL, optic lobe; mcAL, antennal lobe, mechanosensory neuropil. (B) High power view of the boxed region in Fig 5A showing the nuclear label within neurons near the antennal lobe. Arrows point to nuclei that have labeled dots, which are perhaps sites of transcription from the P1 promoter. Bar, 20 µm.

View larger version (41K): In this window In a new window Download PPT slide   Figure 6. Location of labeled groups of neurons in fruw24/fruw27 and wild-type male brain and ventral nerve cord. The sections are selected from a complete series of camera lucida drawings of 20-µm sections through a wild-type animal and represent various horizontal planes through the CNS. These sections were chosen to demonstrate the series of neuronal groups labeled in the CNS of wild-type males by in situ hybridization with probes to the P1 promoter (numbers according to LEE et al. 2000 ; numbers in parentheses labeled according to GOODWIN et al. 2000 ; M. FOSS, G.-H. LEE, E. REYNAUD, L. C. RYNER, J. C. HALL, B. S. BAKER and B. J. TAYLOR, unpublished results). In the fru mutant male (left), neurons (red dots) are present in regions where the neuronal groups are found in wild-type males (red dots, right) but fewer neurons are labeled. The neuronal clusters were labeled in the brain by their relative dorsal-to-ventral location and in the ventral nerve cord by their anterior-to-posterior location. (A) Dorsal section showing labeled neurons in the dorsal posterior part of the brain (1 in both LEE et al. 2000 and GOODWIN et al. 2000 ). (B) Section through the dorsal protocerebrum showing neurons in the lateral group near the optic lobes (10, LEE et al. 2000 ; 2, GOODWIN et al. 2000 ) and neurons from group 1. (C) Section through the protocerebrum showing the location of the three anterior groups (5 and 6, LEE et al. 2000 ; 3–5, GOODWIN et al. 2000 ) distributed along the lateral-to-medial dimension. (D) Ventral section just below the esophagous, which shows the location of the anterior neurons near the mechanosensory antennal lobe (7, LEE et al. 2000 ; 6, GOODWIN et al. 2000 ; Fig 5A). (E) Midway section through the ventral nerve cord showing the location of the abdominal cluster of neurons (a subset of group 20; s20, LEE et al. 2000 ; 9, GOODWIN et al. 2000 ). (F) More ventral section showing the location of the neuronal cluster (a subset of group 17; s17, LEE et al. 2000 ; 7, GOODWIN et al. 2000 ) between the pro- and mesothoracic neuropils. (G) Ventral section showing the location of the mesothoracic neuronal cluster (a subset of group 18, LEE et al. 2000 ; 8, GOODWIN et al. 2000 ).

To determine whether the fru-expressing neurons were still present in these fru mutant animals we hybridized a riboprobe that detects transcripts initiated at the P1 promoter to sections of pharate adult Df(3R)fru^w24/In(3R)fru^w27 and T(3;het)fru^w12/Df(3R)fru^w24 pupae. This probe should detect transcripts in the mutants even though productive transcripts from the P1 promoter are not made due to lesions within the transcription unit downstream of the region homologous to the probe. We examined sections for the six brain clusters and the three male-specific clusters in the ventral nerve cord mentioned above. Neurons were detected in each of these nine regions of the brain and ventral nerve cord in mutant males and the six regions of the brain in females and these clusters were in locations similar to those of the neuronal groups found in wild-type animals (Fig 5A and Fig B, and data not shown; RYNER et al. 1996 ; GOODWIN et al. 2000 ). Thus, at least some neurons that normally express sex-specific fru transcripts are present in these mutant animals, even though no productive sex-specific fru transcripts are being made in these cells (Fig 2B and Fig C). Unlike the expression pattern in wild-type animals, which shows both nuclear and cytoplasmic signals (RYNER et al. 1996 ), in these fru mutants, the label was found only in the nucleus. There was no evidence for cytoplasmic staining, suggesting that the transcripts detected are not transported from the nucleus. The low level of expression in these fru mutant animals is likely why this in situ hybridization signal was not detected in whole-mount preparations (LEE et al. 2000 ). There were fewer neurons found in these new fru mutant animals compared to wild type or to viable fru alleles, such as fru² (data not shown; GOODWIN et al. 2000 ), suggesting that these transcripts either may be less stable or produced less abundantly or that some fru-expressing cells have died or have not differentiated in these fru mutants. In homozygous Df(3R)fru^sat15 third instar larvae labeled with an antisense riboprobe to the P1 promoter, a few labeled neurons were present in the brain, indicating that even in these fru mutant larvae that die early during metamorphosis (see above), at least a few fru-expressing neurons are present (n = 6; data not shown). In addition, these findings suggest that fru-expressing neurons, at least in the major fru-expressing clusters of neurons, do not require the production of intact sex-specific fru transcripts for survival at least up to the pharate adult stage of development.

In contrast to the signal detected by RT-PCR techniques (Fig 2), essentially no signal was found by in situ hybridization using riboprobes generated from the common protein-coding region in Df(3R)fru^w24/In(3R)fru^w27 (male, n = 3, female, n = 3, pharate adult; male, n = 2, female, n = 1, midpupal stage) and T(3;het)fru^w12/In(3R)fru^w27 (male, n = 3, female, n = 3, pharate adult). The in situs to these two genotypes were not distinguishable from those to Df(3R)fru^w24/Df(3R)fru^sat15 (third larval instar, male, n = 3, female, n = 3; data not shown), which is deleted for the common coding region. It may be that the expression of transcripts detected by RT-PCR in Df(3R)fru^w24/In(3R)fru^w27 and T(3;het)fru^w12/In(3R)fru^w27 is below the level of detection by in situ hybridization or that they are expressed in other tissues (see DISCUSSION).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The initial molecular characterizations of fru (RYNER et al. 1996 ; HEINRICHS et al. 1998 ; GOODWIN et al. 2000 ) showed that it produced a complex array of products through the use of at least three promoters and alternative splicing. We undertook the isolation and characterization of additional fru alleles with two major goals in mind: first, to gain insight into the relationship between fru's various functions and the complex array of products produced by fru; second, to characterize in more detail the phenotypic effects of fru mutations to better understand the nature of fru's wild-type functions. The mutations reported here have already contributed significantly to our understanding of fru. Thus a preliminary characterization of several of the new alleles reported here (RYNER et al. 1996 ) showed that fru encoded a sex-nonspecific vital function. In addition, the finding that fru transcripts could still be detected in genotypes generated with these mutations that lacked all previously known fru promoters (P1–3) led to the identification of the P4 promoter (L. C. RYNER, S. F. GOODWIN, T. CARLO, M. FOSS, J. C. HALL, B. J. TAYLOR and B. S. BAKER, unpublished results).

On the relationships between fru's products and functions:
We proposed (RYNER et al. 1996 ) that fru's role in controlling male sexual behavior was through the products of the distal (P1) fru promoter. Briefly, the reasoning was as follows. Genetic analysis (TAYLOR 1992 ; TAYLOR et al. 1994 ) had shown that while all aspects of male somatic sexual differentiation were controlled by the transformer (tra) and transformer-2 (tra-2) sex determination genes, doublesex (dsx), the only known sex determination regulatory gene below tra and tra-2 in the sex determination hierarchy, did not control two aspects of male sexual differentiation (the MOL and male sexual behavior). These results indicated that there was a previously unrecognized branch of the sex determination hierarchy below tra and tra-2 and led to the molecular search for other genes that were direct targets of tra and tra-2. That search identified fru as a direct target of tra and tra-2 in that the transcripts from the P1 fru promoter were sex-specifically alternatively spliced under the control of the TRA and TRA-2 proteins (RYNER et al. 1996 ; HEINRICHS et al. 1998 ). That fru function was known to be required for some aspects of male sexual behavior and the formation of the MOL (GILL 1963 ; HALL 1978 ; GAILEY and HALL 1989 ; GAILEY et al. 1991 ) and that the fru P1 transcripts were expressed in regions of the CNS previously implicated in certain aspects of male sexual behavior (RYNER et al. 1996 ) made a strong case for the products of the P1 fru promoter functioning to control aspects of sexual differentiation in the CNS needed for male sexual behavior and the MOL formation. However compelling that case may seem, it falls short of establishing that the products of the fru P1 promoter have the proposed functions, since there was no demonstration that mutations that impair P1 fru function alone had the expected phenotypes. Recent data consistent with the proposed function of P1 fru products have come from the following findings: (1) the fru¹ allele, which affects male sexual behavior, is an inversion broken 5' to the P1 promoter and alters the spatial pattern of P1 expression in the CNS (and based on its location probably does not affect expression from the other fru promoters; GOODWIN et al. 2000 ); and (2) the fru³ and fru⁴ alleles, which have strong effects on male sexual behavior (VILLELLA et al. 1997 ), result in a substantial reduction in the amount of normal P1 and P2 (but not P3 and P4)-derived transcripts (GOODWIN et al. 2000 ).

Further support for the proposed roles of the P1 fru transcripts is provided by various genotypes involving the new fru alleles reported here. Our findings that males, which express P3 and P4 transcripts, but no P1 or P2 transcripts [i.e., Df(3R)fru^4-40/Df(3R)P14, Df(3R)fru^4-40/T(3;Y)fru^w9, Df(3R)fru^4-40/T(3;het)fru^w12, Df(3R)fru^4-40/In(3R)fru^w27, and Df(3R)fru^4-40/Df(3R)fru^sat15], are sterile and exhibit little or no male-specific reproductive behaviors in single-pair tests strongly support the hypothesis that the products encoded by fru's P1 (and possibly P2) transcripts control male sexual behavior. Flies of some of the above genotypes show reduced general activity and thus one might argue that this contributed to their reduced sexual behavior. However, as quantified in RESULTS, there was in general only a weak correlation between the scores on courtship tests and the test of general activity. Moreover, Df(3R)fru^4-40/Df(3R)P14 males had wild-type levels of general activity and Df(3R)fru^4-40/T(3;Y)fru^w9 and Df(3R)fru^4-40/Df(3R)fru^sat15 males had moderate levels of activity, indicating that for these genotypes the absence of male sexual behavior is unrelated to any general behavioral deficit.

That males having P1-encoded functions either lacking or seriously impaired show as severe decrements in their sexual behavior as males lacking both P1- and P2-encoded functions suggests that fru's role in sexual behavior may be entirely attributed to its P1-encoded functions. Males lacking or severely subnormal for P1 functions are represented by Df(3R)Cha^M5 heterozygous with T93;Y)fru^w9, T(3;het)fru^w12, In(3R)fru^w27, or Df(3R)fru^sat15. In single-pair tests of courtship behavior, males of all these genotypes show essentially no sexual behavior. Moreover, all four types of males showed either normal or moderate levels of general activity. Since a genotype that lacks only P2-encoded functions is not available, we cannot exclude the possibility that P2-encoded functions are also important for male sexual behavior. However, all of the data currently available can be explained by the proposal that fru's P1-encoded functions are responsible for fru's role in male sexual behavior.

The new fru genotypes we analyzed also provide some insight into which fru products are necessary for fru's vital function. With respect to whether P2 encodes a vital function, two aberrations, Df(3R)fru^4-40 and In(3R)fru^w27, both of which lack P1 and P2 transcripts and produce P3 and P4 transcripts, gave completely discordant results when used in complementation tests with other fru alleles that lack fru's vital function. Thus Df(3R)fru^4-40 over Df(3R)fru^sat15 or Df(3R)P14 is fully viable (Table 3C), whereas In(3R)fru^w27 over Df(3R)fru^sat15, Df(3R)Cha^M7, Df(3R)fru^w24, T(3;het)fru^w12, or T(3;Y)fru^w9 is lethal (Table 3D). We believe that this disparity is due to In(3R)fru^w27 being defective in more than just fru's P1 and P2 functions. Our reasoning is as follows. First, the In(3R)fru^w27 chromosome is a complex rearrangement having four cytologically detectable breakpoints (Table 1), one of which is in fru. In(3R)fru^w27 has two molecularly identified lesions in fru, one lesion being in the DNA of the fru² P element on which In(3R)fru^w27 was induced (which is located between P3 and P4; Fig 1B) and the second lesion, presumably the cytologically detectable one, being between P2 and P3 (Fig 1B) and preventing transcripts from going from P1 and P2 but not from P3 and P4 to the common coding region. Second, in viability tests, as well as all the other phenotypic tests we carried out, In(3R)fru^w27 behaves like T(3;Y)fru^w9 and T(3;het)fru^w12, which lack P1, P2, and P3 function. While In(3R)fru^w27 has P3 transcripts detected by RT-PCR, its phenotypic similarity to T(3;Y)fru^w9 and T(3;het)fru^w12 suggests it does not have P3 function. These observations can be reconciled if In(3R)fru^w27 either does not produce P3-derived proteins or expresses them in aberrant temporal or spatial patterns. With respect to the latter possibility note that In(3R)fru^w27 is associated with an inversion broken in fru, which is likely between P2 and P3, and would thus juxtapose new sequences upstream of P3. These new sequences may either lead to abnormal expression of P3 or introduce an ectopic promoter. If this reasoning with respect to In(3R)fru^w27 is correct, and the Df(3R)fru^4-40 results are reflective of the loss of both P1 and P2 functions, then P2-derived products, like those derived from P1, are not needed for viability. Thus fru's vital functions must be carried out by P3- and/or P4-derived products. That P3-encoded functions are required for viability is further suggested by the lethality of a number of mutant combinations that lack P1, P2, and P3 functions (Table 3) but retain P4 function. We could not determine whether the products of the P4 fru promoter encode a vital function, since there is no genotype that lacks P4 function but has P3 function.

In summary, our studies implicate the products of the P1 fru promoter as being responsible for fru's control of male sexual behavior and the products of the P3 fru promoter for carrying out fru's vital function. The functions of the products of the P2 and P4 promoters are currently unclear. However, it needs to be noted that we do not have genotypes that individually remove the functions encoded by P2-, P3-, or P4-derived transcripts. Thus our conclusions as to the functions of P1- and P3-derived transcripts are the simplest ones compatible with our data, but we cannot exclude the possibilities that P2-, P3-, or P4-derived products may play some role in male sexual behavior or that P4-encoded products may also carry out a vital function. Moreover, while our findings that flies lacking P1 and P2 promoter-derived products are viable whereas flies lacking P1-, P2-, and P3-derived products are lethal are most simply compatible with the proposal that P3 products carry out a vital function, these data do not preclude some models in which there is redundancy between the P1, P2, and P3 products in providing fru's vital function.

fru's role in male sexual behavior:
From a detailed analysis of the new fru alleles we showed that wild-type fru function is necessary for the production of nearly all aspects of male sexual behavior. fru mutant males lacking P1-encoded products no longer performed any courtship behavior with either single male or female partners in standard courtship tests (Table 5). The observed failure of males of various fru genotypes to court might reflect a specific defect in sexual behavior, or alternatively it might be a by-product of a general reduction in activity. Although males of several fru genotypes did have reduced levels of short-term activity, regression analysis showed that, overall, there was only a poor correlation between the level of activity and male courtship behavior; this indicates that low activity by these males could at most account for only a small part of the variation in male courtship behavior. Moreover, the reduction in courtship in single-pair tests occurred in some males that had moderate to high levels of activity [e.g., Df(3R)fru^4-40/Df(3R)P14 or Df(3R)fru^4-40/Df(3R)fru^sat15]. Further evidence indicating that reduced courtship with a single partner is a specific defect is shown by a number of fru genotypes in which males did not court a male or female partner, for example, T(3;Y)fru^w9/fru³, but did engage in male-male group courtship as measured by the ChI. This dichotomy in CI vs. ChI shows that males were capable of generating at least some courtship behaviors in one situation but did not express that similar behavior in a different setting. Thus, the elimination of male courtship to both males and females in a subset of fru mutant genotypes reflects a sex-specific courtship effect of the loss of fru⁺ function.

There are two ways these results could be interpreted in terms of the wild-type function of fru. First, since male courtship is a dependent action pattern, with the occurrence of one step in the courtship sequence generally requiring the completion of preceding steps, these results could simply mean that these fru males are blocked in some way prior to the very earliest steps of courtship. Alternatively these results could mean that fru⁺ function is necessary for each step of courtship. The data support the latter alternative. Using several hypomorphic alleles, previous studies showed that fru mutant males courted males and females but were blocked at certain individual steps occurring during courtship. For example, fru³ and fru⁴ males are blocked in the middle of the courtship behavioral sequence such that although they court they do not vibrate their wings to generate the courtship song and do not attempt copulation (VILLELLA et al. 1997 ). By comparison, fru¹ males are blocked only in the penultimate step of attempted copulation (HALL 1978 ; GAILEY and HALL 1989 ; GAILEY et al. 1991 ; VILLELLA et al. 1997 ). These studies provided strong evidence that male-specific fru⁺ function is involved with the specific neuronal circuitry required for the wing vibration needed for courtship song and the abdominal bending for copulation (RYNER et al. 1996 ). Various heteroallelic combinations we tested also show similar defects at intermediate to late stages of courtship. Thus many of the combinations of new fru alleles with fru¹ or fru² showed significant levels of courtship activity as measured by either the CI or the WEI yet were sterile or had greatly reduced fertility due to failures to copulate. Similarly, a number of the fru genotypes we tested failed to produce a courtship song during the reduced period that they were extending their wings. Taken together, these findings suggest that fru functions to establish the potential for essentially all aspects of male sexual behavior.

The exact molecular role that the P1 products play in controlling male-specific behaviors is not known, but they are likely to affect the development and differentiation of particular male-specific neurons and neuronal circuits. Of the clusters of P1-expressing neurons that were identified, we focused on nine different clusters of P1-expressing neurons that were identified by in situ hybridization in late stage pharate male brains and are prominent in sectioned preparations (RYNER et al. 1996 ; GOODWIN et al. 2000 ). Males of a genotype [T(3;het)fru^w12/In(3R)fru ^w27] that did not produce any courtship behavior appear to retain many or most of these neurons in the CNS even though these animals are not producing any functional P1 or P2 transcripts. This finding strongly suggests that the primary role of the P1 transcripts is to direct sex-specific differentiation and not direct the production or facilitate the survival of these neurons.

On single-pair vs. group courtship:
One aspect of the courtship phenotypes of the fru mutant combinations that requires special consideration is what might appear to be somewhat discordant results between single-pair courtship tests (both male-male and male-female) and the male-male group courtship (chaining) test. Of the 10 genotypes that we tested that lacked P1 function (Table 5B and Table 5C), all displayed little or no courtship in single-pair tests, but 6 of these genotypes displayed significant levels of male-male group courtship. In addition, of the 15 genotypes we tested in which one of the new alleles was heterozygous with either fru², fru³, or fru⁴, 12 genotypes displayed little or no courtship in single-pair tests (the exceptions all being combinations involving Df(3R)fru^4-40), yet 13 of these genotypes had significant, often substantial, levels of male-male group courtship. In thinking about these data it is important to recall that although single-pair male-female courtship, single-pair male-male courtship, and male-male group courtship are all referred to as courtship, male-male group courtship is a mutant phenotype, whereas single-pair male-female courtship is the wild-type phenotype in D. melanogaster. Mature males courting each other in pairs essentially never occurs in a sustained manner in this species, so single-pair male-male courtship might also be viewed as a "mutant" phenotype. Thus one might expect a priori that there would not be parallel dependencies of these three different "courtship" phenotypes on fru function. In that regard it is worth recalling that the levels of courtship as measured by the CIs of males of a given genotype with single males or single females are highly correlated. Strikingly, a similar linear regression analysis of the mean CI_m-m and the ChI across all fru genotypes showed a similarly strong positive correlation. These high degrees of concordance suggest that these three behavioral assays are measuring behaviors that are largely equivalent in their fru dependency. The difference between wild type and any of the fru mutants that have been tested to date would be that in the fru mutant situations males as well as females would be seen as appropriate partners for whatever level of courtship males of a particular genotype can attain.

As was just noted for a number of genotypes males showed little or no courtship in single-pair tests but were able to carry out the very first steps of courtship (orientation and following), but not later steps, of courtship in group tests. These differences within a genotype are likely related to the different ways single-pair and group tests are done. In particular, male-male group tests are carried out by placing a group of males together for 3–4 days prior to testing, whereas males used in single-pair courtship tests were stored individually until tested. The richer environment experienced by males used in male-male group courtship tests (more stimulation through dynamic interactions between animals and a longer time for association between animals) is likely the reason for the different levels of courtship seen in single-pair and group tests. The preceding interpretation is suggested by the prior observation that fru², fru³, and fru⁴ homozygous males chained little, if at all, when initially placed together in groups, but that the level of chaining increased over the next couple of days (VILLELLA et al. 1997 ). The finding that fru mutant animals perform some courtship steps under group conditions also shows that the appropriate parts of the brain for at least the very initial steps of courtship are still present and can be activated in these genotypes but that the conditions of the test for pairwise courtship do not lead to the activation of these same centers and the expression of orientation and following.

Another striking aspect of these data is that 6 of the 10 genotypes that completely lack P1 promoter-encoded functions showed significant levels of male-male group courtship. The finding that 4 of these 10 genotypes [T(3;het)fru^w12/Df(3R)Cha^M5, T(3;het)fru^w12/Df(3R)fru^4-40, T(3;het)fru^w12/In(3R)fru^w27, and Df(3R)fru^4-40/Df(3R)P14], as well as the previously characterized Df(3R)Cha^M5/Df(3R)P14 combination (VILLELLA et al. 1997 ), showed no courtship in either single pairs or as groups of males is consistent with the conclusion that it is the products of the P1 fru promoter that establish the potential for male sexual behavior. However, the results from the 6 genotypes that show significant male-male group courtship suggest that other factors are involved that we do not understand. There does not seem to be an obvious genetic explanation for these data, since genotypes that should be identical in terms of their arrays of fru products do not have equivalent phenotypes. For example, in Df(3R)Cha^M5/Df(3R)P14 males all courtship behavior is abolished (VILLELLA et al. 1997 ), while Df(3R)Cha^M5/Df(3R)fru^sat15 males showed no courtship behavior in single-pair tests but had a ChI = 36. Similarly, T(3;Y)fru^w9, T(3;het)fru^w12, and In(3R)fru^w27 over Df(3R)Cha^M5 had ChI's of 19, 5, and 36, respectively, yet our data suggest these should be equivalent in terms of fru function. The origin of these differences is unclear.

fru's role in MOL development:
We suggested that the products of the P1 fru promoter are responsible for both MOL development and male sexual behavior. An obvious question is whether this fru-regulated male-specific muscle has any connection with fru-regulated male-specific courtship behavior. To date, the physiological function of the MOL is unknown. It has been demonstrated that the MOL is not necessary for copulation to occur in this species (GAILEY et al. 1991 ), and, in fact, the males of most Drosophila species do not even develop a MOL (GAILEY et al. 1997 ).

Given the data indicating that the P1-derived male-specific FRU proteins govern MOL development, the recent report (USUI-AOKI et al. 2000 ) that expression of a fru cDNA construct encoding a protein lacking the 101-amino-acid male-specific N terminus could partially rescue the MOL phenotype in fru^sat homozygous males and lead to MOL development in wild-type females was surprising. The fact that a FRU protein can induce MOL development in an otherwise wild-type female provides striking evidence that FRU function is sufficient to trigger MOL development. With respect to the question of which class of FRU proteins controls MOL development in wild type, we note the following. The FRU protein shown by USUI-AOKI et al. (2000) to be capable of producing a MOL is identical to a P1-derived FRU protein, except for the absence of the male-specific 101-amino-acid N terminus. It would not be surprising if there were some overlap in the functionality of these two proteins. The heat-shock and GAL4 expression systems used by USUI-AOKI et al. (2000) are likely to produce FRU protein at significantly higher levels than that protein is present in wild type. Thus the production of MOLs in their experiments may be being brought about by the overexpression of a protein whose function partially overlaps that of P1-encoded proteins. Alternatively, it could be that these proteins are functionally fully synonymous, and these functions are biologically determined by the locations and levels at which they are expressed.

fru's vital function:
Our studies of the phenotypes of lethal fru mutant combinations provided insights into the nature of fru's essential function. Genotypes that lack P3 (and P1 and P2) promoter function [T(3;het)fru^w12/Df(3R)fru^sat15, T(3;het)fru^w12/Df(3R)fru^w24, T(3;het)fru^w12/In(3R)fru^w27, and T(3;het)fru^w12/Df(3R)P14] have lethal phases in the mid- to late pupal period, but in all cases a substantial fraction of these individuals are viable if assisted in emerging from the pupal case. Thus the vital function encoded by P3 is only essential very late in development.

P3-encoded functions appear to be necessary for the differentiation of imaginal-disc derivatives, such as legs and wings. The rare survivors with fru lethal genotypes, such as T(3;het)fru^w12/In(3R)fru^w27, showed defects in their external morphology and had specific sensory bristle organs duplicated. Moreover, imaginal discs frequently failed to evert in adult escapers of genotypes that lacked P3 function. These observations suggest that there is a role for fru in imaginal disc development or differentiation. The gross morphological defects produced in adult derivatives of imaginal discs make it likely that the nearly lethal phenotype of fru genotypes lacking P3 function arises from these defects.

In addition, many of the adult escapers that lacked P3 function had neuroanatomical defects in the motor innervation to the abdominal muscles. These defects in the motorterminals may explain part of the inability of many of these fru mutant genotypes to successfully eclose from the pupal case. All abdominal muscles appear to be innervated, suggesting that the defects in these fru mutants are not due to abnormalities in neuronal pathfinding but rather some feature of synaptogenesis once the motorneuronal axons reached their target muscles. Changes in the branching of motorneurons at the neuromuscular junction in larval Drosophila were described for various mutants that affect neuronal excitability. The number and size of motorneuronal terminals are increased on larval muscles in Shaker and Shaker ether-a-go-go mutant animals, which show increased neuronal excitability, whereas terminals are reduced in number and size in no action potential and paralytic mutants with reduced excitability (BUDNIK et al. 1990 ; JIA et al. 1993 ; reviewed in KESHISHIAN et al. 1996 ). In other respects the development of the nervous system of fru mutant genotypes appeared normal. We did not find any obvious change in the number or patterns of divisions of postembryonic neuroblasts in either a null fru genotype [Df(3R)fru^w24/DF(3R)fru^sat15] or a genotype lacking P1, P2, and P3 functions [Df(3R)fru^w24/In(3R)fru^w27], suggesting that the normal complement of adult-specific neurons is likely produced in the complete absence of fru function. The apparently normal level of postembryonic divisions of neuroblasts in larval CNSs of Df(3R)fru^sat15/Df(3R)fru^w24, which are destined to die within a day, makes it unlikely that the loss of adult neurons is responsible for the death of these animals.

While there is currently not a fru genotype that is null for all fru functions and wild type for neighboring genes, it was possible to place a limit on how extreme a completely null fru genotype might be by examining the lethal phase of mutant combinations [Df(3R)fru^sat15/Df(3R)fru^sat15 and Df(3R)fru^sat15/Df(3R)fru^w24] that were null for fru as well as a small number of adjacent genes. The development of these individuals is arrested in early pupal development, around the time of pupal ecdysis (Table 9). The morphology of the pupal case in these two Df(3R)fru^sat15 genotypes is similar to the phenotypes of animals mutant for the ecdysone receptor and the crooked legs gene (BENDER et al. 1997 ; D'AVINO and THUMMEL 1998 ). However, the above fru genotypes lack not only all fru function but also the function of at least four, and possibly six, neighboring genes. Thus it is not clear if the earlier lethal phase caused by these two genotypes—as compared to fru genotypes that are lacking just P1, P2, and P3 function—is a consequence of the absence of P4 function or the absence of one of the neighboring genes. In any case, these two null genotypes establish that a complete absence of fru function does not result in a lethal phase prior to the beginning of the pupal period.

In summary, our results establish that fru functions sex-specifically in the sex-determination regulatory hierarchy to control male sexual behavior and sex-nonspecifically to control the development of imaginal discs and motorneuronal synapses during development. Perhaps most importantly these results show that just the absence of the products of the P1 fru promoter results in the loss of all normal male sexual behavior and thus provide strong support for the proposal (RYNER et al. 1996 ) that these proteins function to establish the potential for nearly all components of a Drosophila male's sexual behavior.

	ACKNOWLEDGMENTS

We thank R. Levis for p^p[w A] DNA and H. Bellen for synaptotagmin antibody.

Manuscript received September 29, 2000; Accepted for publication May 10, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALBAGLI, O., P. DHORDAIN, C. DEWEINDT, G. LECOCQ, and D. LEPRINCE, 1995  The BTB/POZ domain: a new protein-protein interaction motif common to DNA- and actin-binding proteins. Cell Growth Differ. 6:1193-1198[Abstract].

BAINBRIDGE, S. P. and M. BOWNES, 1981  Staging the metamorphosis of Drosophila melanogaster. J. Embryol. Exp. Morphol. 66:57-80[Medline].

BENDER, M., F. B. IMAM, W. S. TALBOT, B. GANETSKY, and D. S. HOGNESS, 1997  Drosophila ecdysone receptor mutations reveal functional differences among receptor isoforms. Cell 91:777-788[Medline].

BERNSTEIN, A. S., E. K. NEUMANN, and J. C. HALL, 1992  Temporal analysis of tone pulses within the courtship songs of two sibling Drosophila species, their interspecific hybrid, and behavioral mutants of D. melanogaster. J. Insect Behav. 5:15-36.

BUDNIK, V., Y. ZHONG, and C.-F. WU, 1990  Morphological plasticity of motor axons in Drosophila mutants with altered excitability. J. Neurosci. 10:3754-3768[Abstract].

CASTRILLON, D. H., P. GONZY, S. ALEXANDER, R. RAWSON, and C. G. EBERHART et al., 1993  Toward a molecular genetic analysis of spermatogenesis in Drosophila melanogaster: characterization of male-sterile mutants generated by single P-element mutagenesis. Genetics 135:489-505[Abstract].

D'AVINO, P. P. and C. S. THUMMEL, 1998  crooked legs encodes a family of zinc finger proteins required for leg morphogenesis and ecdysone-regulated gene expression during Drosophila metamorphosis. Development 125:1733-1745[Abstract].

FINLEY, K. D., B. J. TAYLOR, M. MILSTEIN, and M. MCKEOWN, 1997  dissatisfaction, a gene involved in sex-specific behavior and neural development of Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 94:913-918[Abstract/Free Full Text].

GAILEY, D. A. and J. C. HALL, 1989  Behavior and cytogenetics of fruitless in Drosophila melanogaster: different courtship defects caused by separate, closely linked lesions. Genetics 121:773-785[Abstract/Free Full Text].

GAILEY, D. A., B. J. TAYLOR, and J. C. HALL, 1991  Elements of the fruitless locus regulate development of the muscle of Lawrence, a male-specific structure in the abdomen of Drosophila melanogaster adults. Development 113:879-890[Abstract].

GAILEY, D. A., S. OHSHIMA, S. SANTIAGO, J. MONTEZ, and A. ARELLANO et al., 1997  The muscle of Lawrence in Drosophila: a case of repeated evolutionary loss. Proc. Natl. Acad. Sci. USA 94:4543-4547[Abstract/Free Full Text].

GILL, K. S., 1963  A mutation causing abnormal courtship and mating behavior in males of Drosophila melanogaster (abstr.) Am. Zool. 3:507.

GOODWIN, S. F., B. J. TAYLOR, A. VILLELLA, M. FOSS, and L. C. RYNER et al., 2000  Aberrant splicing and altered spatial expression patterns in fruitless mutants of Drosophila melanogaster. Genetics 154:725-745[Abstract/Free Full Text].

HALL, J. C., 1978  Courtship among males due to a male-sterile mutation in Drosophila melanogaster. Behav. Genet. 8:125-141[Medline].

HALL, J. C., 1994  The mating of a fly. Science 264:1702-1714[Abstract/Free Full Text].

HAZELRIGG, T., R. LEVIS, and G. M. RUBIN, 1984  Transformation of white locus DNA in Drosophila: dosage compensation, zeste interaction, and position effects. Cell 36:469-481[Medline].

HEINRICHS, V., L. C. RYNER, and B. S. BAKER, 1998  Regulation of sex-specific selection of fruitless 5' splice sites by transformer and transformer-2. Mol. Cell. Biol. 18:450-458[Abstract/Free Full Text].

HU, S., D. FAMBROUGH, J. R. ATASHI, C. S. GOODMAN, and S. T. CREWS, 1995  The Drosophila abrupt gene encodes a BTB-zinc finger regulatory protein that controls the specificity of neuromuscular connections. Genes Dev. 9:2936-2948[Abstract/Free Full Text].

ITO, H., K. FUSJITANI, K. USUI, K. SHIMIZU-NISHIKAWA, and S. TANAKA et al., 1996  Sexual orientation in Drosophila is altered by the satori mutation in the sex-determination gene fruitless that encodes a zinc finger protein with a BTB domain. Proc. Natl. Acad. Sci. USA 93:9687-9692[Abstract/Free Full Text].

JIA, X.-X., M. GORCZYCA, and V. BUDNIK, 1993  Ultrastructure of neuromuscular junctions in Drosophila: comparison of wild type and mutants with increased excitability. J. Neurobiol. 24:1025-1044[Medline].

KESHISHIAN, H., K. BRODIE, A. CHIBA, and M. BATE, 1996  The Drosophila neuromuscular junction: a model system for studying synaptic development and function. Annu. Rev. Neurosci. 19:545-575[Medline].

KULKARNI, S. J. and J. C. HALL, 1987  Behavioral and cytogenetic analysis of the cacophony courtship song mutant and interacting genetic variants in Drosophila melanogaster.. Genetics 115:461-475[Abstract/Free Full Text].

LEE, G., M. FOSS, S. F. GOODWIN, T. CARLO, and B. J. TAYLOR et al., 2000  Spatial, temporal, and sexually dimorphic expression patterns of the fruitless gene in the Drosophila central nervous system. J. Neurobiol. 43:404-426[Medline].

LEE, G., A. VILLELLA, B. J. TAYLOR, and J. C. HALL, 2001  New reproductive anomalies in fruitless-mutant Drosophila males: extreme lengthening of mating durations and infertility associated with defective serotonergic innervation of reproductive organs. J. Neurobiol. 147:121-149.

LEVIS, R., T. T. HAZELRIGG, and G. M. RUBIN, 1985  Separable cis-acting control elements for expression of the white gene of Drosophila. EMBO J. 4:3489-3499[Medline].

LINDSLEY, D. L., and G. G. ZIMM, 1992 The genome of Drosophila melanogaster. Academic Press, San Diego.

LITTLETON, J. T., H. J. BELLEN, and M. S. PERIN, 1993  Expression of synaptotagmin in Drosophila reveals transport and localization of synaptic vesicles to the synapse. Development 118:1077-1088[Abstract].

MOSES, K., M. C. ELLIS, and G. M. RUBIN, 1989  The glass gene encodes a zinc-finger protein required by Drosophila photoreceptor cells. Nature 340:531-536[Medline].

ROBERTSON, H. M., C. R. PRESTON, R. W. PHILIS, D. JOHNSON-SCHILTZ, and W. K. BENZ et al., 1988  A stable genomic source of P-element transposase in Drosophila melanogaster. Genetics 118:461-470[Abstract/Free Full Text].

RYNER, L. C., S. F. GOODWIN, D. H. CASTRILLON, A. ANAND, and A. VILLELLA et al., 1996  Control of male sexual behavior and sexual orientation in Drosophila by the fruitless gene. Cell 87:1079-1089[Medline].

SOKAL, R. R., and F. J. ROHLF, 1995 Biometry, Ed. 3. W. H. Freeman, New York.

SZABAD, J., 1978  Quick preparation of Drosophila for microscopic analysis. Dros. Inf. Serv. 53:1215.

TAYLOR, B. J., 1992  Differentiation of a male-specific muscle in Drosophila melanogaster does not require the sex-determining genes doublesex or intersex. Genetics 132:179-191[Abstract].

TAYLOR, B. J. and L. KNITTEL, 1995  Sex-specific differentiation of a male-specific abdominal muscle, the Muscle of Lawrence, is abnormal in hydroxyurea-treated and in fruitless male flies. Development 121:3079-3088[Abstract].

TAYLOR, B. J. and J. W. TRUMAN, 1992  Commitment of abdominal neuroblasts in Drosophila to a male or female fate is dependent on genes of the sex-determining hierarchy. Development 121:3079-3088.

TAYLOR, B. J., A. VILLELLA, L. C. RYNER, B. S. BAKER, and J. C. HALL, 1994  Behavioral and neurobiological implications of sex-determining factors in Drosophila. Dev. Genet. 14:411-440.

TRUMAN, J. W. and M. BATE, 1989  Spatial and temporal patterns of neurogenesis in the central nervous system of Drosophila melanogaster. Dev. Biol. 125:145-157.

USUI-AOKI, K., H. ITO, K. UI-TEI, K. TAKAHASHI, and T. LUKACSOVICH et al., 2000  Formation of the male-specific muscle in female Drosophila by ectopic fruitless expression. Nat. Cell Biol. 2:500-506[Medline].

VILLELLA, A., D. A. GAILEY, B. BERWALD, S. OHSHIMA, and P. T. BARNES et al., 1997  Extended reproductive roles of the fruitless gene in Drosophila melanogaster revealed by behavioral analysis of new fru mutants. Genetics 147:1107-1130[Abstract].

WHEELER, D. A., W. L. FIELDS, and J. C. HALL, 1988  Spectral analysis of Drosophila courtship songs: D. melanogaster, D. simulans and their interspecific hybrid. Behav. Genet. 18:673-703.

WHEELER, D. A., S. J. KULKARNI, D. A. GAILEY, and J. C. HALL, 1989  Spectral analysis of courtship songs in behavioral mutants of Drosophila melanogaster. Behav. Genet. 19:503-528[Medline].

This article has been cited by other articles:

				R. C. Bertossa, L. van de Zande, and L. W. Beukeboom The Fruitless Gene in Nasonia Displays Complex Sex-Specific Splicing and Contains New Zinc Finger Domains Mol. Biol. Evol., July 1, 2009; 26(7): 1557 - 1569. [Abstract] [Full Text] [PDF]

				B. J. Dickson Wired for Sex: The Neurobiology of Drosophila Mating Decisions Science, November 7, 2008; 322(5903): 904 - 909. [Abstract] [Full Text] [PDF]

				D. A. Gailey, J.-C. Billeter, J. H. Liu, F. Bauzon, J. B. Allendorfer, and S. F. Goodwin Functional Conservation of the fruitless Male Sex-Determination Gene Across 250 Myr of Insect Evolution Mol. Biol. Evol., March 1, 2006; 23(3): 633 - 643. [Abstract] [Full Text] [PDF]

				L. M. Ditch, T. Shirangi, J. L. Pitman, K. L. Latham, K. D. Finley, P. T. Edeen, B. J. Taylor, and M. McKeown Drosophila retained/dead ringer is necessary for neuronal pathfinding, female receptivity and repression of fruitless independent male courtship behaviors Development, January 1, 2005; 132(1): 155 - 164. [Abstract] [Full Text] [PDF]

				I. Ganguly, T. F. C. Mackay, and R. R. H. Anholt Scribble Is Essential for Olfactory Behavior in Drosophila melanogaster Genetics, August 1, 2003; 164(4): 1447 - 1457. [Abstract] [Full Text] [PDF]

				J. C. Hendricks Genetic Models in Applied Physiology: Invited Review: Sleeping flies don't lie: the use of Drosophila melanogaster to study sleep and circadian rhythms J Appl Physiol, April 1, 2003; 94(4): 1660 - 1672. [Abstract] [Full Text] [PDF]

				L. Michaut, S. Flister, M. Neeb, K. P. White, U. Certa, and W. J. Gehring Analysis of the eye developmental pathway in Drosophila using DNA microarrays PNAS, April 1, 2003; 100(7): 4024 - 4029. [Abstract] [Full Text] [PDF]

				H.-J. Song, J.-C. Billeter, E. Reynaud, T. Carlo, E. P. Spana, N. Perrimon, S. F. Goodwin, B. S. Baker, and B. J. Taylor The fruitless Gene Is Required for the Proper Formation of Axonal Tracts in the Embryonic Central Nervous System of Drosophila Genetics, December 1, 2002; 162(4): 1703 - 1724. [Abstract] [Full Text] [PDF]

				B. Dauwalder, S. Tsujimoto, J. Moss, and W. Mattox The Drosophila takeout gene is regulated by the somatic sex-determination pathway and affects male courtship behavior Genes & Dev., November 15, 2002; 16(22): 2879 - 2892. [Abstract] [Full Text] [PDF]

				B. Chan, A. Villella, P. Funes, and J. C. Hall Courtship and Other Behaviors Affected by a Heat-Sensitive, Molecularly Novel Mutation in the cacophony Calcium-Channel Gene of Drosophila Genetics, September 1, 2002; 162(1): 135 - 153. [Abstract] [Full Text] [PDF]