Aberrant Splicing and Altered Spatial Expression Patterns in fruitless Mutants of Drosophila melanogaster

Corresponding author: Stephen F. Goodwin, Division of Molecular Genetics, Anderson College, University of Glasgow, Glasgow G11 6NU, United Kingdom., stephen{at}molgen.gla.ac.uk (E-mail)

Communicating editor: T. SCHÜPBACH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The fruitless (fru) gene functions in Drosophila males to establish the potential for male sexual behaviors. fru encodes a complex set of sex-specific and sex-nonspecific mRNAs through the use of multiple promoters and alternative pre-mRNA processing. The male-specific transcripts produced from the distal (P1) fru promoter are believed to be responsible for its role in specifying sexual behavior and are only expressed in a small fraction of central nervous system (CNS) cells. To understand the molecular etiology of fruitless mutant phenotypes, we compared wild-type and mutant transcription patterns. These experiments revealed that the fru², fru³, fru⁴, and fru^sat mutations, which are due to P-element inserts, alter the pattern of sex-specific and sex-nonspecific fru RNAs. These changes arise in part from the P-element insertions containing splice acceptor sites that create alternative processing pathways. In situ hybridization revealed no alterations in the locations of cells expressing the P1-fru-promoter-derived transcripts in fru², fru³, fru⁴, and fru^sat pharate adults. For the fru¹ mutant (which is due to an inversion breakpoint near the P1 promoter), Northern analyses revealed no significant changes in fru transcript patterns. However, in situ hybridization revealed anomalies in the level and distribution of P1-derived transcripts: in fru¹ males, fewer P1-expressing neurons are found in regions of the dorsal lateral protocerebrum and abdominal ganglion compared to wild-type males. In other regions of the CNS, expression of these transcripts appears normal in fru¹ males. The loss of fruitless expression in these regions likely accounts for the striking courtship abnormalities exhibited by fru¹ males. Thus, we suggest that the mutant phenotypes in fru², fru³, fru⁴, and fru^sat animals are due to a failure to appropriately splice P1 transcripts, whereas the mutant phenotype of fru¹ animals is due to the reduction or absence of P1 transcripts within specific regions of the CNS.

THE fruitless (fru) gene of Drosophila functions at the head of a recently identified branch of the sex-determination hierarchy (RYNER et al. 1996 ) that is required for most, if not all, steps of male courtship behavior (for reviews see HALL 1994 ; TAYLOR et al. 1994 ; COBB and FERVEUR 1996 ; YAMAMOTO et al. 1998 ). In addition to its male-specific function in the sex hierarchy, fru also encodes a sex-nonspecific vital function (RYNER et al. 1996 ).

The fru locus spans at least 140 kb and produces a complex array of transcripts due to the use of four promoters and alternative splicing (Figure 1A and Figure B; ITO et al. 1996 ; RYNER et al. 1996 ; L. C. RYNER, unpublished results). Transcripts from the distal promoter (P1) are alternatively spliced near their 5' termini to generate sex-specific transcripts (RYNER et al. 1996 ; HEINRICHS et al. 1998 ). It is thought that the male-specific P1 transcripts encode fru's sex determination function, whereas the transcripts produced from the more proximal promoters encode fru's vital function(s). Alternative splicing at the extreme 3' end of the fru transcripts leads to the inclusion of one of three mutually exclusive exons that encode alternative pairs of zinc fingers (RYNER et al. 1996 ). Fifteen transcript classes are possible if the five identified 5' ends (two are produced by sex-specific splicing from P1) are combined with all the identified 3' ends; cDNAs corresponding to seven of these classes have been identified (RYNER et al. 1996 ). However, the potential transcript diversity may be more extensive, due to the discovery of additional fru transcripts containing one of several micro-exons (L. C. RYNER, unpublished results). The transcripts from all fru promoters have open reading frames that encode proteins related to the BTB-ZF protein family (cf. HU et al. 1995 ) or Ttk subgroup (cf. ALBAGLI et al. 1995 ); other members of this family function as sequence-specific transcription factors (READ and MANLEY 1992 ; READ et al. 1992 ; VON KALM et al. 1994 ).

View larger version (15K): In this window In a new window Download PPT slide   Figure 1. Genomic map of the fruitless gene. (A) The fru2, fru3, fru4, and frusat P-element insertion sites are indicated by inverted triangles; the position of the fru1 breakpoint is indicated by an arrow. The locus spans ~140 kb. (B) Organization of the fru transcription units. Open boxes represent genomic fragments that hybridize to fru cDNAs; each box may represent more than one exon. The locations of the four promoters (P1–P4) are based on RYNER et al. 1996 and L. C. RYNER (unpublished results). Transcripts from the P1 promoter are alternatively spliced near their 5' termini to generate sex-specific transcript classes. Additional alternative splicing at the 3' end of the gene inserts one of three mutually exclusive exons (L. C. RYNER, S. F. GOODWIN, M. FOSS, B. J. TAYLOR, J. C. HALL, and B. S. BAKER, unpublished results). Whether all alternative 3' ends are used with P2-, P3-, and P4-derived transcripts is unknown. The regions used to generate probes for Northern analyses are indicated by black boxes above the cDNA structures (see MATERIALS AND METHODS): S, 5' sex-specific probe generated by PCR; B, a 645-bp EcoRI genomic fragment containing a cluster of three copies of the 13-nt TRA/TRA-2 repeats contained within this gene; C, BTB domain probe. (C) Structures of aberrant sex and non-sex-specific transcripts produced by fru2, fru3, fru4, and frusat mutants; question marks represent the uncertainty of the cDNAs' 3' ends. (D) The structure of the transposons P{(w+,ry+)A}, P{lwB}, and P{lacZ, ry+}, also known as P{PZ} (HAZELRIGG et al. 1984 ; WILSON et al. 1989 ; MLODZIK and HIROMI 1992 ). The directions of transcription are indicated by arrows, the shaded box represents genomic white sequences (see text), and the solid box represents genomic l(3)S12 sequences.

Phenotypic analyses of fruitless mutants have extended the previously known roles of fru in male courtship behavior (HALL 1994 ; RYNER et al. 1996 ; VILLELLA et al. 1997 ). The courtship actions of wild-type males consist of a series of behaviors: tapping, orientation, following, wing extension, courtship-song production, licking, attempted copulation, and mating (HALL 1994 ; COBB and FERVEUR 1996 ; see Table 1 for details). Because this sequence of behaviors is a dependent series of actions, and the most severely affected fru-mutant types display virtually no courtship, the question arises whether fruitless may only be necessary for an initial step in the courtship sequence. Evidence that the latter interpretation is incorrect, and that instead fru is essential for a number of individual steps in courtship behavior, comes from the phenotypes of less severely affected fru genotypes. These mutants exhibit defects in certain steps of courtship, but they are able to carry out subsequent steps. For example, fru¹, fru², fru³, and fru⁴ homozygotes exhibit defects very early in courtship, in that they cannot discriminate between females and males as appropriate courtship partners, yet they are able to carry out some later steps of courtship, which they direct at both sexes (GAILEY and HALL 1989 ; GAILEY et al. 1991 ; RYNER et al. 1996 ; VILLELLA et al. 1997 ), whereas fru^sat homozygotes have been reported to exhibit exclusively intermale courtship (ITO et al. 1996 ; YAMAMOTO et al. 1996 , YAMAMOTO et al. 1997 , YAMAMOTO et al. 1998 ). During a subsequent courtship step, fru¹ and fru² males produce a courtship song, but it is abnormal (WHEELER et al. 1989 ; VILLELLA et al. 1997 ), while fru³ and fru⁴ males exhibit some wing extensions but fail to generate song pulses during these wing extensions (RYNER et al. 1996 ; VILLELLA et al. 1997 ). These mutants also differ in their ability to carry out the final steps of courtship in that fru² males are fertile, whereas fru¹, fru³, fru⁴, and fru^sat males do not attempt copulation. These observations on the phenotypes of various fru mutants implicate fru in several specific parts of the courtship sequence and not simply the initiation of male sexual behavior.

fruitless exerts its effects on male sexual behavior through its expression in a limited portion of the central nervous system (CNS; RYNER et al. 1996 ). Transcripts from the distal (P1) fru promoter are restricted to a few hundred neurons of the adult CNS. Behavioral analyses of gynandromorphs implicated several regions of the CNS in particular steps of male courtship (HALL 1977 , HALL 1979 ; SCHILCHER and HALL 1979 ), and our initial assessment of cells expressing the P1 fru transcripts (RYNER et al. 1996 ) led to labeling of largely the same neuronal tissues identified in the earlier studies. However, not all regions of the CNS in which the P1 fru promoter is expressed have yet been specifically linked to male courtship behavior. Another fruitless-derived probe, designed to detect fru transcripts produced from all the promoters, showed that most or all neurons in the CNS, as well as many cells in non-neuronal tissues, express this gene (RYNER et al. 1996 ). This more general expression pattern, stemming from the use of one or more of the proximal fru promoters, is likely responsible for fru's vital function (RYNER et al. 1996 ).

Here, we focus on the molecular etiology of the phenotypes produced by fru alleles that affect male sexual behavior, but not fruitless's vital function. These include four P-element mutations (fru², fru³, fru⁴, and fru^sat), which are inserted between the P1 promoter and the common coding region, and the fru¹ mutation, which is due to an inversion breakpoint just distal to the P1 promoter (Figure 1A). Northern analysis shows that the four P-element mutations alter the size of the fru transcripts produced from the promoters located distal to the inserts. These changes arise in part from aberrant splicing into acceptor sites present within the P-element inserts. Not all transcripts produced from the distal promoters are abnormal in these mutants: reverse transcription (RT)-PCR experiments revealed low levels of RNA species spliced in a normal manner. The finding of some wild-type transcripts in the P-element mutations can account for the fact that they differ in the severity of their phenotypes. In situ hybridization to tissue sections revealed that the spatial pattern of expression in the P-element fru mutants was similar to the wild-type male pattern. In contrast, males homozygous for fru¹ make a normal set of sex-specific transcripts, but exhibit an altered spatial pattern of fruitless expression: a subset of the cells that express fru's sex-specific transcripts in wild type do not express these transcripts in fru¹ homozygotes. These data strongly support the suggestion (RYNER et al. 1996 ) that neurons expressing fru's sex-specific transcripts in the CNS control male sexual behavior.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Drosophila strains and culture conditions:
Cultures of Drosophila melanogaster were reared in 12:12-hr light:dark cycles at 25° and 70% relative humidity on a sucrose-cornmeal-yeast medium containing the mold inhibitor Tegosept. Bottles were cleared of parents after 5 days. Males were collected under mild ether anesthetization 0–6 hr after eclosion and then aged individually on unyeasted autoclaved food (see below). Wild-type Canton-S virgin females were collected 0–6 hr after eclosion, grouped in food vials, and then aged for 1–5 days.

fruitless stocks:
fru¹ was balanced with In(3LR)TM3, Sb (TM3). fru² contains a w⁺, ry⁺-marked P element, P{(w⁺, ry⁺)A} (HAZELRIGG et al. 1984 ), inserted at the fruitless locus (GAILEY and HALL 1989 ; GAILEY et al. 1991 ); this strain had been outcrossed to a "Cantonized" white stock for 11 generations (reextracting w⁺ each time) and was subsequently maintained as a homozygous stock (VILLELLA et al. 1997 ). fru³ and fru⁴ arose by insertions of P{lacZ, ry⁺}, also known as P{PZ} within fru (CASTRILLON et al. 1993 ). Both mutations were outcrossed to a ry⁵⁰⁶ stock for four generations, with reextraction of ry⁺, then maintained in MKRS-balanced stocks (VILLELLA et al. 1997 ). fru^sat is caused by a P{lwB} insert at the locus (ITO et al. 1996 ; YAMAMOTO et al. 1996 , YAMAMOTO et al. 1997 ) and was maintained with In(3LR)TM6B, Hu e. The three deletions Df(3R)Cha^M5, Df(3R)P14, and Df(3R)fru^w24 — referred to as Cha^M5, P14, and fru^w24, respectively (GAILEY and HALL 1989 ; RYNER et al. 1996 ) — were used for behavioral observations.

Courtship observations and recordings:
For single-pair observations, individual fru^sat males (7–10 days old) were placed in a recording chamber (cf. VILLELLA and HALL 1996 ; VILLELLA et al. 1997 ) with either another male of the same genotype or a wild-type female (1–5 days old); the fly pairs were video- and audio-recorded for 5–10 min. A courtship index (CI) was calculated for each fru^sat male as described previously (VILLELLA and HALL 1996 ; VILLELLA et al. 1997 ). The CI represents the percentage of time the male exhibited courtship behavior, including all readily observable courtship behaviors: orientation, following, wing extension, and attempted copulation (e.g., VILLELLA et al. 1997 ). For male-pair recordings, a CI was determined only for the male that initiated courtship first for at least 20 sec (VILLELLA et al. 1997 ). A wing-extension index (WEI; VILLELLA et al. 1997 ) was also calculated for each male and represents the fraction of time either wing was held out between 45° and 90° to the body (in wild-type courtship, these wing extensions almost always lead to courtship song). Sounds produced by courting males were analyzed as described previously (cf. VILLELLA et al. 1997 ; see the legend to Table 1). Separate observations, involving certain features of the courtship sequence that are difficult to discern with the naked eye, were performed at x20 (see the legend to Table 1). For this, the flies were placed within the food-containing chambers of a plastic "courtship wheel" (cf. VILLELLA et al. 1997 ); an individual homozygous fru^sat male (5–7 days posteclosion) was placed with a male of the same genotype and age or with a wild-type virgin female (1–5 days old). The pairs were observed during 1- to 2-hr periods for up to 3 successive days (<3 days for test males that performed all three behaviors indicated in the legend to Table 1 on the first or second day).

For male chaining observations, males were collected at eclosion, aged individually for 6–7 days, and then eight males were grouped per food vial (VILLELLA et al. 1997 ). A chaining index (ChI) was determined for each group: the ChI is the percentage of time three or more males courted one another in groups during a 10-min observation period (cf. VILLELLA et al. 1997 ). The ChI was determined on either day 3 or 4 after grouping, because some of the fru male types exhibit more chaining after being together for a few days (VILLELLA et al. 1997 ).

To test male fertility, males were collected at eclosion, grouped (<10 males per food vial), and aged for 5–8 days. Individual males were placed in a food vial with three to five wild-type virgin females. The presence or absence of progeny was scored 7–10 days later. Vials with no progeny and in which the male was dead at the time of scoring were excluded.

Molecular biology:
RNA isolation, purification, and Northern analyses were as described previously (GOODWIN et al. 1997 ). fruitless DNA probes (Figure 1B) were as follows: common-coding region probe (C, which detects all RNAs that have BTB-domain-encoding sequences in common) contains nucleotides from coordinates 2075 to 2422 (with respect to GenBank accession no. U72492; RYNER et al. 1996 ); a distally derived 5' sex-specific exon probe (S), containing nucleotides (nt) from coordinates 188 to 469 (GenBank accession no. U72492); and tra/tra-2-repeat probe (B), a 645-bp EcoRI genomic fragment of the fruitless gene containing a cluster of three copies of a 13-nt repeat (GenBank accession no. U72491; RYNER et al. 1996 ; as described in this citation, these kinds of repeats to which TRA/TRA-2 bind were originally identified in dsx transcripts). Other probes were to white gene sequences [an 11.7-kb EcoRI fragment extending from map positions -5.1 to +6.7 relative to the transcription-start site of the white gene, numbered as in HAZELRIGG et al. 1984 and isolated from the plasmid pP[(w⁺, ry⁺)A]], and to rosy (ry) sequences [a 7.2-kb HindIII genomic fragment isolated from the plasmid pPZ that contains nucleotides from coordinates 6946–14227 (MLODZIK and HIROMI 1992 ); the ry genomic sequence is under GenBank accession no. Y00308].

RT-PCR:
These reactions were performed as described previously (RYNER et al. 1996 ), using either 5 µg of total adult head RNA or poly(A)⁺ RNA and random primers. A total of 10% of the first-strand synthesis was used in a 50-µl PCR, consisting of 50 mM KCl, 10 mM Tris-HCl, pH 8.3, 1.5 mM MgCl₂, 0.01% Triton X-100, 200 µM dATP, 200 µM dCTP, 200 µM dGTP, 200 µM dTTP, 0.2 µM of each primer, and 2 units of Taq polymerase (Promega, Madison, WI). PCR parameters were 3 min at 94°, then 30 cycles of 1 min at 94°, 1 min at 60°, 3 min at 72°, followed by 5 min at 60° and 20 min at 72°. Secondary nested PCRs were as above and contained 2 µl of the primary PCR product as template. The following gene-specific primers were used: fruitless forward primers: fru-1-for (5' GAATTCGAGGACGTGTGACGAT 3'); P1-for-1 (5' TCGCATTACGCGGCCTTGGACT 3'); P2-for-1 (5' TGCTGCAAAAGAACTCAGTCCGC 3'); P3-for-1 (5' TTCGGCACGAGGAACAATCG 3'); P4-for-1 (5' CACACACACACACATATCGAGTTCC 3'); fruitless reverse primer: fru-7A-rev (5' GGAAAATCGTCTCGAAGTACGGAC 3'); white and rosy reverse primers: w-1-rev (5' ATTTGCTGAGCGAAAGCTCCTGG 3'); w-2-rev (5' TCCGTTGATATTCATCACGCCCAC 3'); ry-1-rev (5' CACAAGGGATTTGACAATGCCAGG 3'); ry-2-rev (5' CTACGAGTGGCAAGCAAACTCCAAG 3'); fru⁴ insert primer: fru4-PL (5' CAATGCTTTCCCCCTCTCTTTC 3'); P-element primer: PL (5' GTGTATACTTCGGTAAGCTTCGG 3'); ribosomal-protein 49 (rp49) forward and reverse primer: rp49-rev (5' GTGTATTCCGACCACGTTACA 3'); rp49-for (5' TCCTACCAGCTTCAAGATGAC 3'). Southern analysis and sequencing of PCR products were as described previously (RYNER et al. 1996 ).

Orientation and position of the P-element transposons:
fru³ and fru⁴ are P{PZ} transposons inserted within the fru locus in unknown orientations (Figure 1A and Figure D; CASTRILLON et al. 1993 ; RYNER et al. 1996 ). Genomic DNA flanking the fru³ and fru⁴ PZ-element inserts was cloned by plasmid rescue (MLODZIK and HIROMI 1992 ). Additional sequences from the nonrescuable side of fru³ and fru⁴ were obtained by inverse PCR (TOWER et al. 1993 ). Restriction analysis, Southern blotting, and DNA sequencing established the orientation of a given insert. fru²'s orientation relative to genomic sequences was determined by MOSES et al. 1989 ; the white gene within the fru² transposon is transcribed in the same direction as fruitless, and the rosy gene is transcribed in the opposite direction. fru^sat arose as an insertion of the P{lwB} element close to the site of fru⁴ (Figure 1A and Figure D; WILSON et al. 1989 ; ITO et al. 1996 ; YAMAMOTO et al. 1996 ). The exact position of the fru^sat insertion site was mapped utilizing fru-specific primers derived from the mapping of fru⁴ (see above) in conjunction with P-element-specific primers. Sequencing of the PCR products indicated that the P{lwB} element in fru^sat had inserted 34 bp distally to the fru⁴ insertion site. In fru^sat, the mini-white gene contained within the transposon is transcribed in the same direction as fruitless.

In situ hybridizations to tissue sections:
Flies were cryostat sectioned at 20 µm in the horizontal plane; the sections were transferred to slides, fixed for 1 hr in RNase-free 4% paraformaldehyde, rinsed in RNase-free PBS-Tx (140 mM NaCl, 50 mM phosphate buffer, pH 7.4, and 0.1% Triton X-100), dehydrated through a methanol series (25–100%), and rehydrated with PBS-Tx. Sections were acetylated with acetic anhydride in 0.1 M triethanolamine (pH 8.0) and hybridized overnight at 55° with 15–100 ng of labeled riboprobe in 150 µl hybridization buffer (50% formamide, 5x SSC, 0.01% Tween-20, 0.01% tRNA, and 0.005% RNase-free heparin per slide). The sections were washed in buffer (600 mM NaCl, 1 mM EDTA, pH 8.0, 10 mM Tris, pH 8.0, 50% formamide) at 55° and incubated at 37° with RNaseT1 and RNaseA. For signal visualization, the sections were blocked for 1 hr at 25° in 5% nonfat powdered milk in PBS-Tx, incubated for 1 hr at 25° with antidigoxygenin Fab fragments coupled to alkaline phosphatase (1:1000; Roche Molecular Biochemicals, Indianapolis, IN) in blocking solution and developed with NBT/BCIP (GIBCO BRL, Grand Island, NY) in NTMT (100 mM NaCl, 50 mM MgCl₂, 100 mM Tris, 0.1% Triton X-100, and 0.015% levamisole, pH 9.5). The sections were mounted in 80% glycerol and 1% propyl-gallate.

Standard protocols (Roche Molecular Biochemicals) were used to synthesize riboprobes from fru cDNA and DNA templates [subcloned into pBluescript SK(+) (Stratagene, La Jolla, CA) or pGEM-T Easy (Promega)]. Linearized probes were synthesized using T3, T7, or SP6 polymerase with digoxygenin-coupled nucleotides (DIG RNA Labeling Kit; Roche Molecular Biochemicals). Two riboprobes were made that detect P1-promoter-derived transcripts: probe S1, derived from a genomic clone (coordinates 1–261 with respect to GenBank accession no. U72491), includes sequences from probe S and an extra 340 nt of genomic sequence. Two common coding-region riboprobes were made: probe C1 (coordinates 2783–3607 with respect to GenBank accession no. U72492) and the overlapping probe C (see above). The S and S1 probes gave identical results, as did the C and C1 probes. For the analysis of in situ hybridization patterns, the following numbers of animals were examined: Canton-S males (S1 probe, n = 18; C1 probe, n = 10), females (S1, n = 10; C1, n = 11); fru¹ males (S1, n = 4; C1, n = 2), females (S1, n = 7, C1, n = 2); fru² males (S1, n = 5; C1, n = 8), females (S1, n = 3; C1, n = 3); fru³ males (S1, n = 3; C1, n = 14), females (S1, n = 3; C1, n = 9); fru⁴ males (S1, n = 4; C1, n = 14), females (S1, n = 4; C1, n = 8); and fru^sat males (S1, n = 9; C1, n = 14), females (S1, n = 6; C1, n = 4).

Statistical analyses:
For the courtship data (see legend to Table 1), nonparametric Kruskal-Wallis pairwise comparisons (JMP, Statistics and Graphics Guide, version 3; SAS Institute Inc., Cary, NC) were performed to determine whether there were differences among the CIs or ChIs recorded for fru^sat males, or between mutant and fru⁺-associated CIs. For the in situ hybridizations (see legend to Table 5), the cell counts were analyzed by one-way ANOVA (Statgraphics Version 5.0; Manugistics Inc., Rockville, MD) using 95% Tukey honestly significant difference (HSD) for contrast to compare numbers of labeled neurons between mutant and wild-type animals of the same sex for a particular neuronal group.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

fru^sat behavior:
One objective of these studies was to elucidate the molecular basis for courtship defects caused by viable fru mutations. Because fru^sat has not been as well characterized behaviorally as the other fru mutants (VILLELLA et al. 1997 ), we first quantified the effects of fru^sat on male sexual behaviors. fru^sat males, when paired with either another fru^sat male or a wild-type female, exhibited similar low levels of courtship (Table 1). Hemizygous fru^sat males showed levels of courtship comparable to those of fru^sat homozygotes (Table 1). Thus the extremely low CIs recorded for fru^sat males, compared to the values for wild-type or fru^sat/+ controls (Table 1), are not due to recessive genetic-background effects. Despite this paucity of courtship, sufficient behavior by fru^sat males was observed to reveal that the residual courtship was not normal. During wing extensions, fru^sat males produced only extremely brief sine-song bouts; no song pulses were generated (Table 1 legend). Detailed inspections of fru^sat male behavior revealed that this mutant exhibits tapping and licking, but no attempted copulation (Table 1 legend). fru^sat males, when grouped together, displayed the typical fruitless behavior of courtship chaining (cf. GAILEY and HALL 1989 ; VILLELLA et al. 1997 ); these males displayed numerous wing extensions toward one another when chaining.

fru^sat males were completely sterile when heterozygous with the deletions fru^w24 (n = 20) or P14 (n = 16). However, 1 fru^sat/Cha^M5 male out of 39 was fertile. fru^sat/fru¹ males were also weakly fertile (14/30 males); this proportion is similar to those observed when fru³ or fru⁴ are heterozygous with fru¹ (~65% in both cases; VILLELLA et al. 1997 ).

Origins of normal fruitless transcripts:
When a probe common to all known classes of fru cDNAs (Figure 1, probe C) was hybridized to Northern blots of poly(A)⁺ RNA from sexed wild-type adult heads, three female-specific (9.0, 8.0, and 7.4 kb), three male-specific (7.9, 6.4, and 5.4 kb), and one common (4.4 kb) transcripts were detected (Figure 2A; RYNER et al. 1996 ).

View larger version (14K): In this window In a new window Download PPT slide   Figure 2. Northern blot analysis of fruitless mRNA from sexed head and body tissues of wild-type flies. A total of 5 µg of poly(A)+ RNA was separated in each lane. Hybridization was performed with the following probes (see Figure 1B): (A–C) probes C, B, and S, respectively. Sizes of the transcripts in kilobases are on the left. These blots were subsequently probed with rp49 (O'CONNELL and ROSBASH 1984 ) to control for loading differences among lanes (see bottom of figure).

Sex-specific fru transcripts are generated by the sex-specific usage of alternative 5' splice sites that are 1590 nt apart in the P1-derived pre-mRNA (RYNER et al. 1996 ; Figure 1B). RYNER et al. 1996 suggested that the three male-specific and three female-specific transcripts are generated from P1-derived transcripts by sex-specific splicing at their 5' ends and by sex-nonspecific alternative splicing to exons containing three alternative zinc-finger pairs at their 3' ends. This predicts that the three classes of female-specific transcripts would have common sequences at their 5' ends, as should the three male-specific forms. To test this, we used a fru probe from upstream of the female-specific 5' splice site (Figure 1, probe B) to probe Northern blots of poly(A)⁺ RNA from sexed adult heads. We detected only the three female-specific (9.0, 8.0, and 7.4 kb) transcripts (Figure 2B) seen previously; no signals were observed in male head RNA, even with long autoradiographic exposures. These results are consistent with the interpretation that all three female-specific fru transcripts share common 5' female-specific sequences.

The sequences of the sex-specific fru transcripts revealed that there are no sequences unique to the male transcripts: both male- and female-specific fru transcripts share common sequences from upstream of the male-specific 5' splice site (RYNER et al. 1996 ). We therefore used a probe from this region (Figure 1, probe S) on Northern blots of poly(A)⁺ RNA from sexed wild-type heads (Figure 2C); only the three female sex-specific transcripts (9.0, 8.0, and 7.4 kb) and three male sex-specific transcripts (7.9, 6.4, and 5.4 kb) were observed. This set of results provides strong support for the conclusion that these six sex-specific fru transcripts arise from sex-specific splicing at the 5' end of transcripts produced from the distal (P1) fru promoter (Figure 1).

Molecular etiology of fru mutant phenotypes:
Because the viable fru alleles are due to lesions outside of coding sequences (Figure 1), we suspected that they might have effects that would be detectable at the transcript level. Northern blots hybridized with the common coding region probe (Figure 1, probe C) showed that in the case of the fru¹ mutant, the relative levels of the male- and female-specific transcripts were not appreciably different from those of wild-type controls (Figure 3). Northern blots involving RNA from the four P-element mutants, hybridized with the common coding region probe (C), revealed that sex-specific differences in the pattern of fru expression were no longer detected (Figure 3). Instead, transcripts of ~7.4 and 4.4 kb were detected in both males and females (Figure 3, A–C). In addition to the anomalous 7.4-kb transcript detected in all four mutants, a novel 3.0-kb transcript was observed in fru³, fru⁴, and fru^sat. These results suggest that, in the four P-element mutants, sequences from the distal P1 promoter are no longer sex-specifically spliced to the common-coding region.

View larger version (44K): In this window In a new window Download PPT slide   Figure 3. Northern blot analysis of fruitless sexed head mRNA from wild-type (WT) and fruitless mutants. A total of 5 µg of poly(A)+ RNA was separated in each lane. Hybridizations were performed with the following probes (see Figure 1B): (A–C) probe C (right part of A is a longer exposure), (D–F) probe S; on the left, the transcript sizes are indicated in kilobases. These blots were subsequently probed with rp49 to assess loading. In D, the one high-molecular-weight transcript in fru3 males is of unknown origin; it was not detected on a subsequent blot of RNA derived from an independent RNA sample (data not shown).

To characterize further the effects of these mutations, we rehybridized the Northern blots shown in Figure 3 (A–C) with a probe from the 5' end of P1-derived transcripts (Figure 1, probe S). Again, in the fru¹ mutant, the fru transcript pattern appears to be normal (Figure 3D). In the four P-element mutants, two patterns of transcripts were seen. First, in fru² and fru^sat, there is an abundant 4.0-kb transcript in males and an abundant 6.0-kb transcript in females (Figure 3E and Figure F; longer exposures revealed a faintly hybridizing band at 1.35 kb in fru^sat); none of these corresponds to a transcript size seen in wild type. Second, fru³ and fru⁴ also produce apparently identical arrays of transcripts; these include two major RNA species, a male-specific 2.0- and a 4.0-kb transcript, found in both sexes (Figure 3D). In addition, there are three minor bands common to fru³ and fru⁴ males and two different minor bands common to fru³ and fru⁴ females. Taken together, the above findings indicate that the array of transcripts in fru¹ males and females is normal, but a novel array of P1-derived transcripts is produced in all the P-element mutants.

The finding of a transcript apparently common to the two sexes in fru³ and fru⁴ is surprising, because all other transcripts detected in these mutants, as well as wild-type with this probe, are sex specific. We therefore wondered whether the 4.0-kb transcripts seen in fru³ and fru⁴ males and females are really identical, or whether they were only fortuitously of the same size. We therefore reanalyzed the Northern shown in Figure 3 with a fru probe from upstream of the female-specific 5' splice site (Figure 1, probe B), which in the wild type detected only the three female-specific transcripts (Figure 2B). In the fru³ and fru⁴ mutants, this probe hybridized only to a single female-specific transcript of 4.0 kb; no transcripts were detected in males. These results indicate that the 4.0-kb transcripts observed in fru³ and fru⁴ females and males are two molecularly distinct species.

There are several striking features of these Northern results:

1. P elements inserted at different locations within a 60-kb region of the large fru intron all lead to the production of novel fruitless transcripts.

2. The fru² and fru^sat mutants produce identical arrays of novel fru transcripts despite the fact that these insertions are ca. 40 kb apart (Figure 1).

3. Similarly, the fru³ and fru⁴ mutations also generate identical arrays of novel fru transcripts. A possible basis for the different effects of these two pairs of fru mutants is that fru³ and fru⁴ are due to P-element insertions of the same transposon, P{PZ}, while fru² and fru^sat are due to the insertion of different P-element transposons [P{(w⁺, ry⁺)A} and P{lwB}, respectively] that share some sequences in common.

RT-PCR analysis of the transcripts generated in P-element fru mutants:
We used RT-PCR to investigate whether the four transposon inserts lead to splicing between fru sequences transcribed from a given fru promoter and sequences in the respective downstream transposon. Because potential splice-acceptor sites within the inserted elements depend on the orientation of the transposon relative to fru, we first determined the orientations of the fru³ and fru⁴ inserts (see MATERIALS AND METHODS). The PZ elements in fru³ and fru⁴ are oriented such that the rosy⁺ marker gene is expressed from the same strand as fru. In the case of fru² and fru^sat, the P element inserted in these mutants carries part of the white gene, including its first intron and the splice sites that flank it. We confirmed previous reports that these two elements are inserted such that the sense strands of fru and the white gene in the P elements are the same (cf. MOSES et al. 1989 ; ITO et al. 1996 ).

The positions of the fru promoters relative to the P-element insertions (Figure 1A) suggest which transcripts might be affected in each mutant. For example, sequences from fru^sat, which is located between P2 and P3 (Figure 1C), would be expected to be included in transcripts originating from the P1 and P2, but not the P3 and P4 promoters, and, thus, might alter the processing of P1- and P2-derived transcripts. Because the Northern results led us to expect the effects of fru² and fru^sat to be similar to one another, and different from those of fru³ and fru⁴, we describe results from these two pairs of mutants separately.

To determine whether there are transcripts being produced in fru² and fru^sat that fuse fru sequences with white sequences from within these transposons, RNA was isolated from these mutants and subjected to RT-PCR; the primer-set combinations represented each fru promoter (P1–P4) as well as nested white-gene-specific primers (see MATERIALS AND METHODS). PCR products were sequenced directly. The nucleotide sequences of such products showed that fru P1 transcript sequences are spliced to those of white in both mutants in each sex (Figure 4A). These splices utilize the normal fru⁺ male- and female-specific 5' splice donor sites, which are spliced to a common acceptor site >70 kb downstream in wild-type P1 transcripts. However, in fru² and fru^sat mutants, these 5' splice sites are joined to the normal acceptor site in the first intron of the white gene of the P{(w⁺, ry⁺)A} and P{lwB) inserts, respectively (PEPLING and MOUNT 1990 ). Conceptual translation of these mRNAs showed that FRU amino acids are fused out of frame with white sequences (Figure 4A).

View larger version (48K): In this window In a new window Download PPT slide   Figure 4. Nucleotide and predicted amino-acid sequences from the truncated P1–P3 transcripts produced by fruitless transposon-insert mutants. fru 5'-end sequences are underlined (GenBank accession no. U72492; L. C. RYNER, S. F. GOODWIN, M. FOSS, T. CARLO, A. VILLELLA, B. J. TAYLOR, J. C. HALL and B. S. BAKER, unpublished results), stop codons are marked with asterisks, and fru 5' splice sites are joined to the normal acceptor site in the first intron of the white gene of the P{(w+, ry+) A} and P{lwB} inserts, respectively, at positions 80–81 (coordinates with respect to GenBank accession no. X51749; PEPLING and MOUNT 1990 ). (A) Nucleotide and predicted amino-acid sequences from the hybrid P1 transcripts produced by fru2 and frusat mutant males. The positions of the fusions between fru and white sequences are indicated by vertical arrows (out-of-frame white-encoded amino acids are italicized). (B) Nucleotide and predicted amino-acid sequences from the hybrid P2 transcripts produced by fru2 and frusat males. (C) Nucleotide and predicted amino-acid sequences from the hybrid P3 transcript produced by fru2 males. (D) Nucleotide and predicted amino-acid sequences from the hybrid P1 transcripts from fru3 and fru4 males; the position of the fusion between the fru and l(3)S12 sequences is indicated by a vertical arrow (out-of-frame l(3)S12-encoded amino acids are italicized). (E) Nucleotide and predicted amino-acid sequences for the hybrid P2 transcript from fru3 and fru4 mutants.

We also detected aberrant transcripts derived from the other fru promoters located distal to these inserts. In fru² mutants, transcripts from P2 and P3 promoters are spliced via the normal fru 5' splice donor sites (L. C. RYNER and S. F. GOODWIN, unpublished results) to the normal acceptor site in the first intron of the white gene of the P{(w⁺, ry⁺)A} (Figure 4B and Figure C), whereas P4 transcripts are normal. In fru^sat mutants, as predicted from its enhancer trap location between P1 and P2, P2 transcripts are spliced to white sequences from the P{lwB} insert (Figure 4B), whereas P3 and P4 transcripts appear to be normal.

Both the fru³ and fru⁴ inserts are located between the P2 and P3 promoters and, thus, might be expected to affect transcripts from the P1 and P2 promoters. HOROWITZ and BERG 1995 previously showed that the HindIII fragment of the rosy gene that was used to make the PZ element (MLODZIK and HIROMI 1992 ) contains a portion of another gene, l(3)S12 (located adjacent to rosy), including a portion of its intron, the splice-acceptor site, and some coding and transcription-termination sequences. To ascertain whether aberrant transcripts are produced in fru³ and fru⁴ as a result of splicing of fru transcripts into the rosy or l(3)S12 genes, we employed RT-PCR analysis with gene-specific primer sets representing each fru promoter (P1–P4), with primers based on the rosy and l(3)S12 gene sequences (see MATERIALS AND METHODS). The nucleotide sequences of the PCR products demonstrated that sequences from the fru P1 promoter are spliced to those of l(3)S12 in both mutants (Figure 4D). This splice utilizes the normal fru⁺ male- and female-specific 5' splice donor sites in males and females, respectively, which are joined to a sequence in l(3)S12 previously identified as a splice-acceptor site (HOROWITZ and BERG 1995 ). Conceptual translation of these fusion cDNAs showed that fru sequences are fused out of frame with the l(3)S12 ORF (Figure 4D).

We also examined the effects of the inserts in fru³ and fru⁴ on the processing of transcripts from the other fru promoters. Aberrant transcripts derived from the P2 fru promoter were detected in both fru³ and fru⁴. Transcripts from the P2 promoter are spliced via the normal fru 5' splice donor site (L. C. RYNER and S. F. GOODWIN, unpublished results) to the previously identified 3' splice site in l(3)S12 of the PZ element present in these two mutants (Figure 4E). In both fru³ and fru⁴, processing of transcripts from the P3 and P4 promoters is normal, as predicted from the locations of these promoters downstream of the insertion sites in these two mutants. Thus, in fru³ and fru⁴, as in fru² and fru^sat, processing of all transcripts initiating upstream of a P-element insertion site is disrupted, whereas processing of transcripts initiated from downstream promoters is normal.

The findings that sequences in all four transposons lead to nonproductive splicing of fru transcripts offers a first-order explanation for how transposon inserts generate fru-mutant phenotypes. However, what is not accounted for by these findings is that these four insertion mutations differ in the severity of their phenotypic effects. One possible explanation for these phenotypic differences is that there are varying levels of normal fru transcripts generated from the P1 promoter in these mutants. Although normal fru P1 transcripts were not detected in our Northern analysis (see above), it is possible that such transcripts are present, but at levels below those detectable by Northern analysis (~1–10 pg of transcript; cf. SAGERSTROM and SIVE 1996 ). We therefore used the more sensitive technique of RT-PCR (detection limit is 0.01–0.1 pg of transcript; SAGERSTROM and SIVE 1996 ) to ask whether normal fruitless transcripts are generated in these mutants.

RT-PCR experiments utilizing total RNA failed to detect wild-type fru transcripts from the P1 or P2 promoters in the P-element mutants, in contrast to wild-type controls (data not shown; see MATERIALS AND METHODS). In addition, P4 sex-nonspecific transcripts were detectable in all the mutants tested; P3 transcripts were detected in fru³, fru⁴, and fru^sat males, but not in fru² males. This experiment was repeated twice using different RNA isolates, and the RNA's integrity in the RT-PCR reaction was checked using primers derived from the rp49 gene (see MATERIALS AND METHODS); in all samples, we observed amplification of rp49 sequences. In a further effort to identify normal fru transcripts in these mutants, we used head poly(A)⁺ mRNA isolated from wild-type male and female heads (separately)—and from sexed heads of fru², fru³, fru⁴, and fru^sat flies—for RT-PCR analysis. Sequencing of PCR products was used to confirm their identity. Although these RT-PCR experiments were not quantitative, we were able to detect the normal array of spliced fru products from all four promoters in all the P-element mutants (data not shown).

CNS expression patterns in fru mutants:
An important feature of P1-derived transcripts is that they are expressed in several hundred neurons in the CNS of adults (RYNER et al. 1996 ). We analyzed the spatial pattern of P1 expression in wild type in more detail and, at this higher level of resolution, examined whether any of the viable fru alleles alter the gene's spatial expression pattern.

In situ hybridization to P1 transcripts (using the S1 probe; Figure 1) revealed that P1 expression in wild-type pharate adults was largely localized to nine neuronal clusters in the brain and ventral nerve cord (Figure 5A and Figure B). Labeled neurons were considered part of a group if the signal-containing cells were adjacent to or within a few cell diameters of other similarly sized labeled cells. By these criteria, we identified six neuronal groups, ranging from 10 to 50 neurons per group, which are found in similar locations within male and female brains (Figure 5A), and three neuronal groups in the ventral nerve cord that are male specific (Figure 5B). All nine groups are present as bilateral pairs.

View larger version (111K): In this window In a new window Download PPT slide   Figure 5. Distributions of neurons expressing sex-specific fruitless transcripts in the nervous system of wild-type and fru-mutant males and females. These schematic representations of the CNS are from camera-lucida drawings of a complete series of 20-µm horizontal sections through a wild-type, late-stage pharate adult; the same diagrams were used as templates for expression patterns in the fru1, fru2, and fru4 mutants. The positions of labeled neurons are based on in situ hybridizations performed with the S1 probe to serial sections of a single individual of a particular genotype (see Figure 1; MATERIALS AND METHODS). The positions of labeled cells were mapped onto the appropriate section, in relation to morphological features visualized with Normarski optics. Neurons in males are indicated by red dots on the left side and in females by blue dots on the right side of the section. The locations of the groups of labeled neurons described in the text are indicated by numbers on the male half of the sections. Because the expression patterns of labeled neurons were similar among fru3, fru4, and frusat, only those sites for a fru4 brain are shown. In the brain series in Figure 5A, the first section is the most dorsal (D) one, and the last section in the series is the most ventral (V). On each section through the brain, anterior is the bottom side of the section and posterior is at the top. For the sections through the ventral nerve cord in Figure 5B, the dorsalmost section is to the left and the ventralmost section is to the right; anterior (A) is at the top and posterior (P) is at the bottom. (A) Distribution of neurons labeled with the S1 probe in the brains of wild-type, fru1, fru2, and fru4 (from left to right). Arrows indicate regions where fewer neurons associated with groups 3, 4, and 5 were found in fru1 male brains. More heavily labeled neurons are present in all sections of fru1 female brains when compared to fru2, fru3, and fru4 females (open arrows point to regions where the additional neurons were most visible). The increase in the number of labeled neurons in frusat females was similar to that in fru1 females. (B) Distribution of neurons in the ventral nerve cords of wild type, fru1, fru2, and fru4 (from left to right). Solid arrows indicate regions in the abdominal ganglion (group 9) in which labeled neurons were not detected in fru1, as well as cells in the thoracic ganglion (group 7), where fewer labeled neurons were found in fru1 male ventral nerve cords compared to the other genotypes. More neurons were labeled in all sections of fru1 female ventral nerve cords (examples indicated by open arrows) compared to females of all other genotypes.

In the brain, the locations of the six groups are as follows: 1, a large group in the dorsal posterior protocerebrum, medial and ventral to the mushroom bodies; 2, a lateral group in the protocerebrum, anterior to the medullary division of the optic lobes; 3–5, three anterior groups in the protocerebrum which are subdivided into lateral (3), intermediate (4), and medial (5) sets; and 6, one anterior group near the mechanosensory part of the antennal lobe (Figure 5A; cf. RYNER et al. 1996 ). In the ventral nerve cord, the three groups of male-specific fru P1-expressing neurons are as follows: 7, a lateral group between the prothoracic and mesothoracic neuromeres and ventral to the wing neuromere; 8, a ventro-medial group in the mesothoracic ganglion; and 9, a ventral group in the abdominal ganglion (Figure 5B). In addition to these groups of cells, both sexes have a small population of labeled neurons that are found as singletons or are too widely separated from other labeled cells to recognize them as belonging to a group (cf. RYNER et al. 1996 ).

In contrast to the restricted pattern of expression of P1 fru transcripts, a probe (C1) that detects the protein-coding region common to all fru transcripts (Figure 1) revealed expression in virtually all neurons in the CNS and in several other tissues (cf. RYNER et al. 1996 ). Most cells in the CNS have a relatively low level of expression, whereas a small number of neurons display markedly higher levels of fru expression. By their location and number, the subsets of neurons labeled by the C1 probe that have relatively high levels of fru expression appear to correspond to the nine groups of neurons detected by the S1 probe (cf. RYNER et al. 1996 ; B. J. TAYLOR, unpublished results).

To determine whether there were changes in the cellular pattern of fru expression in the five viable fruitless mutants, we carried out analogous in situ hybridizations with both the S1 and C1 probes. Only the presence and relative size of the aforementioned nine groups of labeled neurons were analyzed, because these groups of fru-expressing cells are distinct and could be unambiguously identified in tissue sections.

Examination of the expression pattern of P1-derived transcripts in fru¹ males revealed distinct groups of labeled neurons in only four of the nine regions where cells expressing these transcripts are found in wild-type males. The four groups of neurons detected were those in the dorsal posterior protocerebrum (1), the optic lobe (2), the antennal lobe (Figure 6A and Figure B, group 6), and the ventro-medial mesothoracic groups (Figure 5A, group 8). The numbers of labeled neurons in the antennal lobe and dorsal posterior groups in fru¹ males were not significantly different from wild type (Table 2).

View larger version (111K): In this window In a new window Download PPT slide   Figure 6. Photomicrographs of in situ hybridizations in horizontal sections of fruitless mutants. (A–E) P1 promoter expression pattern in the mechanosensory part of the antennal lobe in males (cf. section 11 from the top in Figure 5A). Arrows in high-magnification views (B–E) point to labeled group-6 neurons (only cells on the right side are indicated, but both left and right sides are shown). (A) Low-magnification view of a fru1 male head. (B) High-magnification view of the boxed region in A with labeled neurons (arrows) in the antennal lobe. (C) High-magnification view of a fru2 male. (D) High-magnification view of a fru3 male. (E) High-magnification view of a fru4 male. (F–J) P1 promoter expression pattern in anterior lateral protocerebrum in males (cf. seventh section in Figure 5A). Arrows point to the labeled neurons in groups 3 and 4 in the dorsal anterior part of the protocerebrum. (F) Low-magnification view of the head of a fru1 male. (G) High-magnification view of the boxed region in A; only one side of the protocerebrum is shown in the high-magnification image, and the midline is to the left in G–J. (H) High-magnification view of a fru2 male. (I) High-magnification view of a fru3 male. (J) High-magnification view of a fru4 male. (K–N) P1 promoter expression pattern in the thoracic ganglia of males (cf. ninth section from the left in Figure 5B). Arrows point to labeled neurons in thoracic neuronal group 7 in L and M. (K) Low-magnification view of the thoracic ganglia in a fru1 male. (L) High-power view of the boxed region in K. (M) High-magnification view of thoracic neurons in a fru2 male. (N) High-magnification view of thoracic neurons in a fru3 male. (O–R) P1 promoter expression pattern in the abdominal ganglion (cf. seventh section from the left in Figure 5B). Arrows point to labeled neurons in the abdominal neuronal group 9 in Q and R. (O) Low-magnification view of the ventral nerve cord of a fru1 male. (P) High-magnification view of the boxed region in O of the abdominal ganglion, which has no labeled neurons. (Q) High-magnification view of the abdominal cluster of neurons in a fru2 male. (R) High-magnification view of the abdominal cluster of neurons in a fru3 male. (S–U) P1 promoter expression pattern in female fru CNS. (S) High-magnification view of labeled mechanosensory antennal lobe neurons (group 6; section 11 in Figure 5A); arrows point to neurons on both sides of the midline. (T) High-magnification view of labeled neurons (group 3; section 7 in Figure 5A) in the anterior lateral protocerebrum. (U) High-magnification view of the neurons labeled in the thoracic ganglia, showing relatively limited numbers of strongly labeled neurons and many nearby cells with moderate-to-light levels of staining; there are more labeled neurons in this region in fru1 and frusat mutant females than in wild-type, fru2, fru3, or fru4 mutant females. (X–Z) Labeled neurons in the brain and ventral nerve cord of fru2 mutant males, detected by the common coding probe (C1). (X) Low-magnification view of the mechanosensory antennal lobe (group 6; cf. section 12 in Figure 5A); the arrow points to the location of neurons with a relatively high level of expression; by their locations and relative number, these heavily labeled neurons are likely to be the same ones detected by the S1 probe; all other neuronal cell bodies in the section also show a low level of fru expression with this probe. (Y) High-magnification view of the boxed region indicated by arrow in A. (Z) High-magnification view of the thoracic group 7 (cf. ninth section in Figure 5B, showing that the male-specific groups of neurons in this region are also detected by probe C1 and show a relatively high level of staining). In all of these images, anterior is to the top. The scale bars for the low-magnification views are 100 µm and for the high-magnification ones are 20 µm.

The other five groups of cells detected in wild-type males were very difficult to detect in fru¹ males. In the anterior protocerebrum, the normal pattern for three groups of neurons (3–5) was not observed (Figure 6F and Figure G). In the dorsal anterior protocerebral region, only about one-quarter of the expected number of labeled neurons was detected (Table 2). These labeled neurons were distributed throughout the medial, intermediate, and lateral subdivisions, suggesting that fru¹ leads to a reduced number of cells expressing P1 transcripts in all three regions. It was not possible to identify definitively the remaining labeled neurons as belonging to one of the three anterior protocerebral groups. Likewise, labeled neurons were difficult to detect in two of the male-specific neuronal groups in the ventral nerve cord of fru¹. A small cluster of neurons, ~15% of the expected number, was found in the ventral area, between the prothoracic and mesothoracic neuromeres (Figure 6K and Figure L; Table 2). By contrast, almost no labeled cells were found in the ventral abdominal region, where group 9 neurons are observed in wild-type males (Figure 6O; Table 2). The reduced numbers of cells expressing P1 transcripts in fru¹ males may be caused by a reduction in transcription from the P1 promoter, instability of these transcripts, or the loss of neurons that normally express these transcripts.

When a probe (C1) common to all fruitless transcripts was used, all neurons in fru¹ male CNSs were labeled at a low level, comparable to what is seen in the wild-type CNS. In addition, subsets of neurons showed relatively high levels of fru expression. Neurons with such heavy labeling were detected in the dorsal posterior protocerebrum (1), optic lobe (2), antennal lobe (6), and mesothoracic groups (8); these are the same regions in which the neurons expressing the sex-specific transcripts are abundant in fru¹ males (see above). In fru¹, fewer-than-normal numbers of heavily labeled neurons were detected with the C1 probe in the anterior protocerebrum (3–5) as well as within the thoracic (7) and the abdominal ganglion (9) groups (data not shown). These regions are the same as those in which fewer or no labeled neurons were found by in situ hybridization with the S1 probe (Figure 5A and Figure B, and Figure 6A, Figure B, Figure F, Figure G, Figure K, Figure L, Figure O).

We also used in situ hybridization to determine whether the spatial expression of P1-derived transcripts was affected in the transposon mutants. In all four mutants, the S1 probe detected groups of labeled neurons in the nine locations where P1-expressing neurons are found in wild-type males (see the schematic representations in Figure 5A and Figure B; because no differences among the expression patterns for fru², fru³, fru⁴, and fru^sat were detected, only the locations of labeled neurons for a fru⁴ brain and thoracic/abdominal ganglia are shown). For example, labeled neurons were found in the antennal lobe (Figure 6, C–E, group 6), the dorsal anterior protocerebrum (Figure 6, H–J, groups 3 and 4), the thoracic ganglion (Figure 6M and Figure N, group 7), and the abdominal ganglion (Figure 6Q and Figure R, group 9). The numbers of neurons labeled in these groups were similar to those found in wild-type males (Table 2). However, there was an overall increase in the number of heavily labeled neurons in the brains of fru^sat males compared to the other fru mutant or wild-type males (Table 4). In wild-type males, cells labeled with the S1 probe often had darkly stained dots within the nucleus, as well as cytoplasmic staining (cf. RYNER et al. 1996 ). In a similar fashion, in all the P-element-mutant males, neurons labeled with the S1 probe often had very darkly stained dots in the nucleus in addition to weak cytoplasmic staining (Figure 6, C–E, H–J, M, N, Q and R).

In situ hybridization to the four P-element mutants with the C1 probe detected heavily labeled cells only in fru² males. These neurons were in the same positions as the nine groups labeled by the S1 probe (e.g., group 6, cf. Figure 6C and Figure Y; group 7, cf. Figure 6Z and Figure M). These neurons had the same darkly stained dots in the nucleus, as was seen with the S1 probe (see above), in addition to having a higher level of cytoplasmic expression than the surrounding neurons (Figure 6X and Figure Y). Thus, the expression pattern in fru² males appears to be similar to that of wild-type males.

In contrast, in the CNS of fru³, fru⁴, and fru^sat males, no heavily expressing neurons were detected with the C1 probe, although there was a general low level of CNS staining. In these sections, only ~10 neurons in total had evidence of nuclear dots. These findings suggest that detectable levels of transcripts containing normal P1-derived transcripts are not found in the CNS neurons of fru³, fru⁴, and fru^sat males. Because a population of heavily labeled neurons was detected using the S1 probe in these mutants, the fru-expressing neurons are present, but it does not appear that intact P1 transcripts can be detected in these cells by in situ hybridization. This finding is in accord with the RT-PCR results presented above, in which it was necessary to use poly(A)⁺ RNA from heads to detect normally spliced P1-derived transcripts in these mutants.

Using both the S1 and C1 probes, we also examined the effects of all five fru mutations on fruitless expression in females. In the four P-element mutants, the cellular distributions and expression levels of transcripts from the P1 promoter (Figure 5A and Figure B, S1 probe; Table 3) were comparable to those of wild-type females (data not shown). With the C1 probe, only fru² females showed heavily expressing neurons; these neurons were in the same locations as the cells detected by the S1 probe, suggesting that the same neurons are labeled with both probes (cf. RYNER et al. 1996 ). In contrast, in the CNS of fru³, fru⁴, and fru^sat females, no heavily expressing neurons were detected with the C1 probe (data not shown). This suggests that the levels of intact P1 transcripts are also below the level of detection by in situ hybridization in the neurons of fru³, fru⁴, and fru^sat females. Because a population of heavily labeled neurons is detected using the S1 probe, the fru-expressing neurons are present in these females, as they are in males of these genotypes.

Examination of the distribution of P1-derived transcripts (S1 probe) in fru¹ and fru^sat females revealed a substantial change in the distribution of labeled neurons. In fru¹ and fru^sat females, the overall staining intensity and number of labeled neurons were increased in the CNS compared to wild-type females (Figure 5A and Figure B, and Figure 6, S–U; Table 3). The normal pattern of six groups of labeled neurons was still present in fru¹ and fru^sat female brains, and the numbers of labeled neurons were not significantly different from those in wild-type or other P-element mutants (Figure 5A; Table 3). The increase in the number of labeled neurons appeared in regions where labeled neurons are not found in wild-type males and females, e.g., in ventral regions of the thoracic ganglion (Figure 5B and Figure 6U). This increased expression of transcripts from the P1 promoter was confined to the CNS, because no expression was detected in the non-neural tissues of fru¹ and fru^sat females (data not shown). The difference between fru¹ and wild-type females in the numbers of heavily labeled neurons in the ventral thoracic CNS was statistically significant; although the numbers in the brain were also larger than normal, this difference was not statistically significant (Table 4).

When the C1 probe was used on fru¹ females, we observed heavily labeled neurons as well as widespread low-level neuronal expression in the CNS. As was found with the S1 probe, more neurons gave strong signals with the C1 probe in fru¹ than in wild-type females.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHOD