The fruitless Gene Is Required for the Proper Formation of Axonal Tracts in the Embryonic Central Nervous System of Drosophila

Corresponding author: Barbara J. Taylor, 3029 Cordley Hall, Oregon State University, Corvallis, OR 97331-2914., taylorb{at}bcc.orst.edu (E-mail)

Communicating editor: T. SCHÜPBACH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The fruitless (fru) gene in Drosophila melanogaster is a multifunctional gene that has sex-specific functions in the regulation of male sexual behavior and sex-nonspecific functions affecting adult viability and external morphology. While much attention has focused on fru's sex-specific roles, less is known about its sex-nonspecific functions. We have examined fru's sex-nonspecific role in embryonic neural development. fru transcripts from sex-nonspecific promoters are expressed beginning at the earliest stages of neurogenesis, and Fru proteins are present in both neurons and glia. In embryos that lack most or all fru function, FasII- and BP102-positive axons have defasciculation defects and grow along abnormal pathways in the CNS. These defects in axonal projections in fru mutants were rescued by the expression of specific UAS-fru transgenes under the control of a pan-neuronal scabrous-GAL4 driver. Our results suggest that one of fru's sex-nonspecific roles is to regulate the pathfinding ability of axons in the embryonic CNS.

THE fruitless (fru) gene has a prominent role in developing the potential for male sexual behavior as well as adult viability and normal external morphology (ITO et al. 1996 ; RYNER et al. 1996 ; GOODWIN et al. 2000 ; USUI-AOKI et al. 2000 ; ANAND et al. 2001 ; LEE and HALL 2001 ; LEE et al. 2001 ). fru's functional complexity is reflected in the structural complexity of the locus. The fru gene encodes a large set of sex-specific and sex-nonspecific transcripts generated by differential promoter usage and alternative splicing at the 5' and 3' ends (ITO et al. 1996 ; RYNER et al. 1996 ; GOODWIN et al. 2000 ; USUI-AOKI et al. 2000 ). fru transcripts from four different promoters, P1, P2, P3, and P4, encode closely related BTB/POZ (Broad complex, Tramtrack, and Bric-a-brac/ Poxvirus and Zinc finger)-Zn finger (ZnF) proteins, which likely act as transcription factors (ITO et al. 1996 ; RYNER et al. 1996 ; GOODWIN et al. 2000 ; LEE et al. 2000 ; USUI-AOKI et al. 2000 ; ANAND et al. 2001 ).

The male-specific behavioral functions of fru depend on transcripts produced from the P1 promoter that are sex-specifically spliced (ITO et al. 1996 ; RYNER et al. 1996 ; GOODWIN et al. 2000 ; USUI-AOKI et al. 2000 ; ANAND et al. 2001 ). In males a default 5' acceptor splice site is used, whereas in females, the Transformer (Tra) and Transformer-2 (Tra-2) proteins direct splicing to a second downstream acceptor site (RYNER et al. 1996 ; HEINRICHS et al. 1998 ; USUI-AOKI et al. 2000 ). In both sexes, alternative splicing at the 3' end leads to the generation of three types of P1 transcripts, each with a different pair of ZnF domains (GOODWIN et al. 2000 ; USUI-AOKI et al. 2000 ). A consequence of the sex-specific splicing of P1 primary transcripts is the generation of a set of P1 transcripts in males, which upon translation produce a family of proteins with an amino-terminal extension preceding the BTB/POZ domain and a set of P1 transcripts in females, which are not translated (ITO et al. 1996 ; RYNER et al. 1996 ; GOODWIN et al. 2000 ; LEE et al. 2000 ; USUI-AOKI et al. 2000 ). The male-specific P1 transcripts are translated into male-specific Fru proteins (Fru^M), which are expressed in nearly 2000 neurons in the pupal and adult central nervous system (CNS) and play an important role in male sexual behavior and in the formation of a male-specific abdominal muscle known as the muscle of Lawrence (ITO et al. 1996 ; RYNER et al. 1996 ; GOODWIN et al. 2000 ; LEE et al. 2000 ; USUI-AOKI et al. 2000 ; ANAND et al. 2001 ). Recently, it has been shown that the Fru^M proteins are required for the presence of the serotonin neurotransmitter in a set of abdominal neurons that innervate the masculine internal genitalia and mediate aspects of male fertility (LEE and HALL 2001 ; LEE et al. 2001 ).

The products of the P2, P3, and P4 fru promoters differ from those of the P1 promoter in their spatial patterns of expression, their protein coding sequences, and their functions (RYNER et al. 1996 ; GOODWIN et al. 2000 ; ANAND et al. 2001 ). Transcripts from the P2, P3, and P4 promoters encode a small amino-terminal extension of a few amino acids preceding the common BTB/POZ domain compared to the 101 amino-terminal motif found in the P1-derived male-specific proteins (S. F. GOODWIN, L. C. RYNER, T. CARLO, M. FOSS, J. C. HALL, B. J. TAYLOR and B. S. BAKER, unpublished results). In situ hybridization using a probe from the common coding region of fru, which detects all or most of the known fru transcripts, shows that fru transcripts from non-P1 promoters are ubiquitously expressed in the adult CNS and in particular subsets of nonneuronal tissues during larval and pupal development of both adult male and female flies (RYNER et al. 1996 ; GOODWIN et al. 2000 ; LEE et al. 2000 ). Anti-Fru antibodies, which recognize the common region of Fru proteins, confirm that these fru RNAs are translated in both sexes in the CNS and nonneuronal tissues (LEE et al. 2000 ). Genetic analysis of fru mutants shows that mutants lacking the P3 and/or P4 transcript classes die at the pupal stage, suggesting that the expression of a subset of fru transcripts is essential for adult viability (RYNER et al. 1996 ; ANAND et al. 2001 ). In some fru mutant genotypes, adult escapers show a variety of defective external phenotypes, indicating that these transcripts likely have some function in the development of these adult structures (ANAND et al. 2001 ).

To gain a better understanding of the role of fru's P2, P3, and P4 transcripts in development, we determined the expression pattern of fru RNAs and proteins in the embryonic CNS. In this article, we show that fru RNAs and proteins are produced in a dynamic pattern during embryogenesis and are widely expressed in the developing embryonic CNS. Neuronal precursors as well as neurons and glial cells express Fru proteins. The embryonic CNS is an important model system for defining how specific genes govern neuronal identity and the process of axonal pathfinding necessary for the formation of proper neuronal connections (reviewed in GOODMAN and DOE 1993 ; GOODMAN 1996 ; TEAR 1999 ). The main axonal tracts in the ventral nerve cord of the fruit fly embryonic CNS consist of two bilaterally symmetrical longitudinal connectives with a pair of commissures, anterior and posterior, that cross the midline in each segment (GOODMAN and DOE 1993 ). The proper formation of this longitudinal and commissural axon scaffold is complex and involves the directed growth of axonal processes from interneurons and motoneurons in specific fascicles within the CNS (e.g., reviewed in TEAR 1999 and RUSCH and VAN VACTOR 2000 ). Diffusible molecules secreted from cells along the midline are one of the mechanisms involved in attracting or repulsing specific axons (GUTHRIE 1999 ; reviewed in TEAR 1999 ; BROSE and TESSIER-LAVIGNE 2000 ). Axons, as they grow through the CNS, selectively fasciculate and defasciculate to find their appropriate synaptic targets (TEAR 1999 ; RUSCH and VAN VACTOR 2000 ).

Analyses of fru mutants, in which specific subsets of transcripts are disrupted or in which no fru transcripts are made, revealed that Fasciclin II (FasII)-positive axons had abnormal trajectories, as did axons producing other longitudinal and commissural tracts in the CNS. Our findings suggest that fru transcripts are needed in the embryonic CNS for the formation of the wild-type pattern of axonal tracts. The defects in the FasII-positive and BP102-positive axonal tracts found in fru mutants were rescued by the expression of specific fru isoforms under the control of the pan-neuronal sca-GAL4 driver, which has a pattern of expression similar to that of Fru proteins. In contrast, these mutant phenotypes were not rescued by fru transgene expression driven by the elav-GAL4 driver, which is expressed exclusively in postmitotic neurons. The neurons that pioneer the FasII fascicles express markers appropriate to their wild-type identity in fru mutants. Our analysis suggests that fru's primary function during neurogenesis is the regulation of fasciculation and defasciculation processes involved in the growth of neuronal axons along their wild-type pathway through the CNS.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly stocks and crosses:
Canton-S flies were used as the wild-type genotype. The following fru mutant alleles were used: fru³; fru⁴; fru^sat; In(3R)fru¹ (fru¹); Df(3R)Cha^M5 (Cha^M5); Df(3R)fru^w24 (fru^w24); In(3R) fru^w27 (fru^w27); Df(3R)fru^sat15 (fru^sat15); Df(3R)fru^4-40 (fru^4-40); In(3R)fru^w12 (fru^w12); Df(3R)P14 (P14) (CASTRILLON et al. 1993 ; ITO et al. 1996 ; RYNER et al. 1996 ; ANAND et al. 2001 ). fru mutant alleles were maintained over either TM6B, Tb, Hu or TM3, Sb, P[ftz-lacZ]. To label midline glial cells in fru mutants (see below), the P element in the enhancer-trap line AA142 (SCHOLZ et al. 1997 ; Bloomington Stock Center) was recombined onto the fru^sat15 and fru^AJ96u3 chromosomes. Fly stocks were maintained at room temperature on an agar, sucrose, cornmeal, and yeast medium supplemented with 0.1% nipagin (p-hydroxybenzoic acid methyl ester; Sigma, St. Louis) for mold inhibition.

A deficiency, Df(3L)XDI98 (65A02-65E1), was used to examine the phenotypes associated with the non-fru inversion breakpoint, at 65C-D, of the fru^w12 chromosome. fru^w12/Df(3L) XDI98 mutants were fully viable, fertile, and morphologically normal adults (n = 10 male; n = 10 female). Deficiency heterozygous males did not show courtship chaining behaviors, a typical fru mutant phenotype (n = 20, 10 males each group). Due to the similarity of phenotypes, a spread mutant allele (sprd⁰⁵²⁸⁴; SPRADLING et al. 1999 ; Bloomington Stock Center) was used in complementation tests with various fru mutations. There were no adult sprd⁰⁵²⁸⁴/fru^w24 survivors, indicating that large fru deficiencies uncover the sprd locus (0/36 control siblings). However, sprd⁰⁵²⁸⁴ trans-heterozygotes with other fru alleles were fully viable with a normal wing phenotype (sprd⁰⁵²⁸⁴/fru^4-40, n = 61; sprd⁰⁵²⁸⁴/fru^w12, n = 30; sprd⁰⁵²⁸⁴/fru^AJ96u3, n = 24) and did not form male-male courtship chains (sprd⁰⁵²⁸⁴/fru^4-40, n = 20; sprd⁰⁵²⁸⁴/fru^w12, n = 8; sprd⁰⁵²⁸⁴/fru^AJ96u3, n = 9).

Generation and molecular characterization of the fru^AJ96u3 deficiency:
The AJ96w⁺ P element (SPANA and DOE 1996 ) was mapped by isolating flanking genomic sequences using inverse PCR followed by sequencing (YEO et al. 1995 ). These query sequences were used for BLAST homology searches against the Drosophila genome and mapped to genomic sequences that include the fru locus (accession no. AE003722; FLYBASE 1999 ). To create new fru mutations, excisions of the AJ96w⁺ P element were induced by standard techniques (ROBERTSON et al. 1988 ). One lethal excision out of 56 w^- revertant lines was recovered. The molecular limits of the fru^AJ96u3 deficiency were determined by a combination of Southern blot analysis and genomic PCR using oligo primers obtained from plasmid subclones from across the fru region and flanking sequences (data not shown). The exact molecular locations of the deficiency breakpoints were determined by using oligo primers, flanking these sequences, to PCR amplify fragments that contain sequences around the breakpoints. These PCR fragments were sequenced and compared to the published sequences for this region.

Lethal phase, phenotypic, fertility, and behavioral analysis of fru^AJ96u3 mutant animals:
To characterize the fru^AJ96u3 mutation, trans-heterozygotes between fru^AJ96u3 and other fru alleles were examined. In those crosses in which adult fru trans-heterozygotes did not survive, the pupal stage at which the fru mutant combination died was determined from aged collections of white prepupae and by staging based on BAINBRIDGE and BOWNES 1981 . For genotypes that reached late stages of pupal development, the pupal case was dissected to allow the animals to emerge.

For sterility tests, virgin males were collected a few hours after eclosion and housed as groups of 8–10 males in food vials. After 4 days, individual males were mated with 2 or 3 Canton-S virgin females and the vials were examined after 7 days for the presence of larvae and/or pupae. To determine whether males would form courtship chains (VILLELLA et al. 1997 ), males were collected after eclosion and aged alone for 3–4 days. Eight males of the same genotype were then put together in a food vial and observed in the late afternoon or early evening for the presence of male courtship chains with >3 males courting during a 1-hr period over 3–4 days.

Embryonic immunohistochemistry:
Timed embryo collections were staged by morphological criteria (CAMPOS-ORTEGA and HARTENSTEIN 1997 ) and prepared for immunohistochemistry according to PATEL 1994 . The following primary antibodies were used (except where noted, these antibodies were a gift from N. Patel) to label neurons and glial cells: anti-Fasciclin II (1D4, 1:5; GRENNINGLOH et al. 1991 ), mab22C10 (1:200; FUJITA et al. 1982 ), anti-Elav (9F8, 1:30; O'NEILL et al. 1994 ), mabBP102 (1:20; SEEGER et al. 1993 ), anti-Repo (1:100; a gift of A. Travers; HALTER et al. 1995 ), anti-Engrailed/Invected (4D9, 1:10; PATEL et al. 1989 ), and anti-Odd-skipped (1:200; a gift from J. Skeath; SPANA et al. 1995 ). To label fru-positive cells, rat anti-Fru^COM (1:500; LEE et al. 2000 ), rat anti-Fru^BTB' (1:500; this study, see below), rat anti-Fru^A' (1:500; this study), and rat anti-Fru^C' (1:500; this study) were used. In all experiments, embryos were labeled with anti ß-galactosidase antibodies (1:10,000; Cappel, Durham, NC) to distinguish fru mutant embryos (ß-galactosidase negative) from control sibling embryos (ß-galactosidase positive).

For detection of primary antibodies, we used secondary antibodies, which were directly conjugated with horseradish peroxidase (Jackson ImmunoResearch Laboratories, West Grove, PA), with alkaline phosphatase (AP; Jackson ImmunoResearch), or with the fluorochromes Alexa-594, Alexa-488, or Alexa-395 (Molecular Probes, Eugene, OR). For anti-Repo labeling, a biotinylated secondary antibody was used followed by incubation with the ABC reagent (Vector Laboratories, Burlingame, CA). For some double-label experiments, colorimetric visualization of diaminobenzidine (Sigma) was nickel enhanced (PATEL 1994 ). Enzymatic processing of alkaline phosphatase used NBT/X-phosphatase (Boehringer Mannheim, Indianapolis) in AP reaction buffer (0.1 M NaCl, 0.1 M Tris-HCl, pH 9.5, 0.05 M MgCl₂, 0.1% Tween 20; PATEL 1994 ). To improve the signal obtained from the anti-Fru antibodies, we used the tyramide-based signal amplification kit (New England Nuclear, Boston) prior to staining for AP.

Labeled whole-mount and filleted embryos were viewed and photographed with a Sony DKC-5000 digital camera under differential interference contrast optics, using an Olympus Vanox-TX microscope. For certain double- or triple-labeling experiments, fluorescently labeled embryos were viewed on a TCS-Leica confocal microscope. Composite images were assembled in Adobe Photoshop 5.0.

In situ hybridization of embryos:
Single-strand antisense and sense riboprobes for in situ hybridization were prepared as previously described (RYNER et al. 1996 ; GOODWIN et al. 2000 ). The following antisense riboprobes were used: the common coding region probe (antisense-Com, nucleotides (nt) 2785–3612, GenBank accession no. U72492), the BTB domain region probe (nt 2075–2422, GenBank accession no. U72492), the P1-promoter-derived sex-specific probe (probe P1.S, nt 160,236–159,918, accession no. AE003722), the P2 5' end probe (nt 121,168–121,101 plus 120,967–120,936; AE0037222.2), the P3 5' end probe (nt 95,033–94,958, accession no. AE003722), the P4 5' end probes (nt 61,666–61,268, accession no. AE003722), and the Zn finger motif probes A (nt 1743–2294, accession no. D84437), B (nt 37,721–37,479, accession no. AE003722), and C (nt 3872–4114, GenBank accession no. U72492).

The protocols for fixation and in situ hybridization were according to BROADUS and DOE 1995 except that RNase treatment (1 mg/ml) was included after hybridization to minimize nonspecific binding of the riboprobe. To visualize the signal, embryos were incubated in dilute anti-digoxigenin-AP (1:2000) and reacted for AP (BROADUS and DOE 1995 ). Embryos were mounted in 70% glycerol and viewed as either whole-mount or dissected preparations.

Generation of transformants:
Six different UAS-fru constructs were made from fru cDNAs subcloned in pBluescript KS (RYNER et al. 1996 ; S. F. GOODWIN, L. C. RYNER, T. CARLO, M. FOSS, J. C. HALL, B. J. TAYLOR and B. S. BAKER unpublished results). These constructs all have a minimal 5'-untranslated region (UTR) and were subcloned into a pUAST vector (BRAND and PERRIMON 1993 ; VAN ROESSEL and BRAND 2000 ). The UAS-fruA, UAS-fruB, and UAS-fruC constructs all started at the same 5' EcoRI site (128,354, accession no. AE003722.2) and included identical 5'-UTR, BTB, and common fru coding sequences but different 3' end sequences. The UAS-fruA construct contains a 3-kb EcoRI fragment of fru cDNA no. 7 (L. C. RYNER, personal communication; the A 3' end terminates at nt 40563, accession no. AE003722.2). The UAS-fruB construct contains a 4.5-kb EcoRI-XbaI fragment from a fru cDNA no. 25 (L. C. RYNER, personal communication; the B 3' terminates at nt 36684, accession no. AE003722.2). The UAS-fruC construct contains a 3.6-kb EcoRI-KpnI fragment of the female cDNA no. 1 (GenBank accession no. U72492, nt 1763–5409). The UAS-fruMA, UAS-fruMB, or UAS-fruMC constructs all have the same 5' sequences, which encode the 101 male-specific amino terminus and the BTB domain derived from fru male cDNA nos. 5-19 (L. C. RYNER, personal communication). This cDNA was truncated at the 3' end and subcloned into pMartini (a gift from N. Brown) as a SacI-PvuI fragment. The final constructs were generated via a three-way ligation into the pUAST vector. The 5' sequences (1.2-kb EcoRI-PvuI fragment) were ligated to a 1.948-kb PvuI-EcoRI fragment (A 3' end; fru cDNA no. 7; L. C. RYNER, personal communication), a 1.98-kb PvuI-HindIII fragment (B' end; fru cDNA no. 25; L. C. RYNER, personal communication), or a 1.98-kb PvuI-KpnI (C 3' end; nt 2824–5409, GenBank accession no. U72492). These constructs would encode Fru proteins like those from P4 transcripts and similar to those from P2 and P3 transcripts, which have N-terminal extensions of a few more amino acids than P4 Fru proteins (S. F. GOODWIN, L. C. RYNER, T. CARLO, M. FOSS, J. C. HALL, B. J. TAYLOR and B. S. BAKER, unpublished results).

The pUAST vectors also contained a mini-white reporter gene and the final transgene constructs were introduced into the Df(1)yw parental strain by germline transformation as described in RUBIN and SPRADLING 1982 . Transgene constructs (300 µg/ml) were coinjected with the helper plasmid pUCHS2-3 (100 µg/ml). The chromosomal location of 3–10 transformant lines was determined. If a transformant line contained two insertions on different chromosomes, the transgenes were segregated and treated as independent lines. Each UAS-fru construct was crossed into a fru^W12 or fru^sat15 mutant background and balanced over TM3, Sb, and P[ftz-lacZ]. At least two independent lines for each construct were tested and analyzed.

To express these various UAS-fru constructs in the CNS, lines containing scabrous-GAL4 (sca-GAL4) and embryonic lethal abnormal vision-GAL4 C155 (elav-GAL4; LIN and GOODMAN 1994 ; Bloomington Stock Center) transgenes were crossed into a fru^W12 or fru^sat15 mutant background. The scabrous sequences in the transgenic construct drive pan-neuronal expression of GAL4 from neuroblasts through neurons (KLAES et al. 1994 ; see below) and the elav sequences in this transgenic construct drive expression in postmitotic neurons (LIN and GOODMAN 1994 ). Crosses between a GAL4 line and a UAS-fru line generated the fru mutant embryos in which one of the fru isoforms was expressed in the CNS and peripheral nervous system (PNS). To distinguish fru mutant embryos from control siblings, all embryos were labeled with anti-ß-galactosidase (1:10,000; Cappel), which permitted unlabeled fru mutant embryos to be distinguished from labeled control siblings, expressing the lacZ marker.

Generation of Fru antibodies:
For the glutathione S-transferase (GST) fusion constructs, coding regions for the fru BTB domain and alternative ZnF A and C (Fig 1) were generated by PCR. Oligonucleotide pairs containing sites for in-frame directional cloning in the pGEX-4T-1 vector (Pharmacia, Piscataway, NJ) were designed for BTB and each unique ZnF sequence. DNA was amplified from fru cDNA clones (primer sequences for BTB domain are BTB-1-For, GGG GGA ATT CAT GGA CCA GCA ATT CTG CTT 3'; BTB-115-Rev, GGG GGC TCG AGC TAG TTG TTA TCT GTG AGA 3'; those for A form ZnF domain are A-For, 5' CCG GAA TTC CAG CAG CGC CCG CCA CC 3'; A-Rev, 5' GCC GCT CGA GCG GGA TGG GCT GCA CTT GGG C; and those for C form ZnF domain are C-For, 5' CCG GAA TTC CGC GTC AAG TGT TTT AAC ATT AAG C 3'; C-Rev, 5' CCG CTC GAG GTT TGC TTG ATT CTT GGT TAC TTA G 3'), digested with EcoRI and XhoI enzymes, and cloned into the polylinker of the pGEX-4T-1 vector. Individual recombinant clones were validated by sequencing.

View larger version (28K): In this window In a new window Download PPT slide   Figure 1. The molecular map of the fru locus. The insertion sites of the P elements, fru2, fru3, fru4, frusat, and AJ96w+, are indicated by triangles (ITO et al. 1996 ; RYNER et al. 1996 ; GOODWIN et al. 2000 ; ANAND et al. 2001 ; this article). The AJ96w+ P element is located 3 kb from the 3' end of the fru locus (this article). The fru breakpoints of the inversion alleles fru1 and fruw12 are shown on the genomic map, but the inversion allele fruw27 is more complex with the relevant break somewhere between P2 and P3, since no P2 transcripts are detected by reverse transcriptase-PCR (ANAND et al. 2001 ). The limits of relevant fru deficiencies, including the Df(3R)fruAJ96u3 described in this article, are delineated with thick black lines to represent deleted sequences and dashed lines to represent breakpoints mapped to the relevant restriction fragment (ANAND et al. 2001 ; this article). fru exons are positioned on the genomic map. Four promoters (P1–P4) are distributed throughout the locus and there are three alternative 3' ends (A, B, and C). Each 3' end encodes a different pair of ZnF domains (GOODWIN et al. 2000 ; USUI-AOKI et al. 2000 ; S. F. GOODWIN, L. C. RYNER, T. CARLO, M. FOSS, J. C. HALL, B. J. TAYLOR and B. S. BAKER, unpublished results). The P1 promoter produces primary transcripts that are spliced in females specifically by the TRA and TRA-2 proteins binding to three repeats located in the S exon. These transcripts also contain the BTB domain and other common exons (C1–5) and are spliced alternatively to one of the three different 3' ends (RYNER et al. 1996 ; GOODWIN et al. 2000 ; USUI-AOKI et al. 2000 ). Transcripts derived from the other three promoters, P2, P3, and P4, contain the BTB and common exons but it is not known whether these transcripts utilize all of the possible alternative 3' ends. Therefore, the full extent of transcript complexity from these promoters is not known (RYNER et al. 1996 ; GOODWIN et al. 2000 ; USUI-AOKI et al. 2000 ; ANAND et al. 2001 ; S. F. GOODWIN, L. C. RYNER, T. CARLO, M. FOSS, J. C. HALL, B. J. TAYLOR and B. S. BAKER, unpublished results).

Fusion proteins were purified according to SMITH 1993 . These fusion proteins were SDS-PAGE purified and not proteolytically cleaved from the GST so that the whole GST-ZINC FINGER or GST-BTB peptides were used as the immunogen. These materials were injected into rats using a 77-day protocol. In brief, 750 µg of purified protein was injected into each animal with complete Freund's adjuvant, followed by three 750-µg immunizations (boosters) using incomplete Freund's adjuvant. Serum was collected by exsanguinations at the end of the protocol. Titer and specificity of the antibodies were assayed by Western blots of the recombinant protein (data not shown). Harvested polyclonal antisera against the fru BTB domain and the A and C forms of the ZnF domain hereafter are named as anti-Fru^BTB', anti-Fru^A', and anti-Fru^C', respectively.

Statistics:
The frequency of defective embryos in various fru mutant and wild-type embryos was analyzed statistically by one-way ANOVA (SAS program version 6.12; SAS Institute) and post hoc analyzed by Tukey HSD comparisons or by two-sample t-test (SAS program version 6.12).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation and characterization of a new fru null mutation:
The AJ96w⁺ P element, which labels a pair of midline neurons in each segment of the embryonic CNS, was mapped molecularly to 3 kb downstream of the 3' end of the fru locus (Fig 1; MATERIALS AND METHODS; SPANA and DOE 1996 ). Homozygous AJ96w⁺ males did not have a fru mutant phenotype as they were fertile (31/32 males, each paired with two or three virgin Canton-S females) and failed to form male-male courtship chains (six tests of 8 males each; total n = 48).

We mobilized this P element and generated a deficiency line, Df(3R)AJ96^U3 (fru^AJ96u3). This deletion extends from sequences within the AJ96w⁺ P element to sequences between the P3 and P4 promoters, thus deleting all fru common coding sequences (Fig 1; see MATERIALS AND METHODS). As the smallest available deficiency of the fru locus, it appears to remove fru and only one other putative open reading frame located between the 3' end of the fru locus and the AJ96w⁺ P element (29,144–27,188 bp, accession no. AE003722; FLYBASE 1999 ). Phenotypic analysis confirmed that fru^AJ96u3 is a fru null allele. fru^AJ96u3 trans-heterozygotes with hypomorphic fru alleles were fully viable (Table 1A) but were sterile (0/18 fertile fru^4-40/fru^AJ96u3 males; 0/18 fertile fru⁴/fru^AJ96u3 males). fru trans-heterozygotes with strong fru alleles, such as fru^w27/fru^AJ96u3, survived to late pupal stages and often would emerge if dissected out of the pupal case, but had striking visible external defects, such as uninflated wings or malformed leg joints (Table 1C and Table 1D; see also ANAND et al. 2001 ). By comparison, fru^AJ96u3 trans-heterozygotes with fru null alleles, such as fru^sat15, survived to early pupal stages, dying at or just after pupal ecdysis, the same lethal phase of other known fru nulls (Table 1B; ANAND et al. 2001 ). To address the possibility that some phenotypes of these fru escapers might be due to loss of function in the nearby spread gene (sprd, map location 91A5-6; SPRADLING et al. 1999 ), we performed complementation testing with various fru alleles and deficiencies. Our results show that these fru mutations fully complement this sprd mutation, supporting the attribution of the lethality and visible mutant phenotypes to the loss of fru function (see MATERIALS AND METHODS; ANAND et al. 2001 ).

Temporal and spatial distribution of fru mRNA during embryogenesis:
To determine the spatial and temporal distribution of embyonic fru transcripts, we performed in situ hybridizations with antisense-BTB and antisense-Com riboprobes, which detect most or all fru transcripts (RYNER et al. 1996 ; GOODWIN et al. 2000 ; LEE et al. 2000 ). We found that fru mRNAs are expressed in a dynamic temporal and spatial pattern from the beginning of embryogenesis until stage 16 (Fig 2; Table 2A). fru transcripts are uniformly distributed in very early embryos and become incorporated into segregating pole cells [stages (st) 1–5; Fig 2A]. At the start of gastrulation (st 6), heavily labeled cells are found in the ventral and cephalic furrows. In slightly older embryos (st 7–9; Fig 2B) the most prominent distribution of fru transcripts is found in the developing CNS within mesectodermal and ventral neuroectodermal cells. Transcripts become localized to delaminating neuroblasts (Fig 2B and Fig G) and after stage 10, fru transcripts are detected in medial but not in lateral neuroblasts (Fig 2H). Following expression in the progeny of neuroblasts, the level of fru expression in the CNS continues to decline until becoming undetectable at stage 16 (Table 2A; Fig 2I). Thus, fru transcripts are expressed throughout the development and early differentiation of the CNS but become undetectable at later stages. Cells in some non-CNS tissues, the amnioserosa and tracheal placodes, also expressed fru transcripts (st 9–11; Table 2A), but no in situ hybridization signal was detected in other tissues, such as the PNS or body wall muscles.

View larger version (132K): In this window In a new window Download PPT slide   Figure 2. The temporal and spatial distribution of fru mRNA and Fru protein. In situ hybridization was carried out with digoxigenin-labeled antisense fruBTB riboprobes and AP histochemistry (A–C and G–I). Fru protein distribution was determined by immunohistochemical localization of anti-FruBTB' followed by AP histochemistry (D–F and J–L). In G–L, embryos were also labeled with anti-Engrailed antibody. (A) Stage 4 wild-type embryo. fru mRNA is found in pole cells (arrowhead). (B) Stage 8 wild-type embryo. Mesectodermal cells (arrow) are strongly labeled for fru mRNA. (C) Stage 11 fru null (fruw24/fruw24) embryo. No fru mRNA was detected in any cell types at this stage. (D) Stage 4 wild-type embryo. Fru protein is found in pole cells (arrowhead). (E) Stage 9 wild-type embryo. Cephalic regions (white asterisk) and two rows of mesectodermal cells (arrow; compare with B) express Fru protein. (F) Stage 14 fru null (fruw24/fruw24) embryo. No Fru protein was detected in any cell types at this stage. (G) Late stage 9 wild-type embryo. fru mRNA is found in three columns of neuroblasts: S1 neuroblasts (l, lateral column and m, medial column) and S2 neuroblasts (i, intermediate column). Engrailed-positive epidermal cells are present in a different focal plane. (H) Stage 10 wild-type embryo. fru mRNA appears to be in most GMCs (white arrow) and is also more strongly expressed in the anterior half of the segment. (I) Stage 11 wild-type embryo. fru mRNA is present but with less intensity than found in earlier stages. (J) Late stage 9 wild-type embryo. Fru protein is found in three columns of neuroblasts: SI neuroblasts (l, lateral column and m, medial column) and SII neuroblasts (i, intermediate column). (K) Stage 10 wild-type embryo. Fru protein is detected throughout the CNS, appears to be in most or all GMCs (white arrows), and is also more strongly expressed in the anterior half of the segment. (L) Stage 11 wild-type embryo. Fru protein is expressed in a large number of cells in the nervous system. There is slightly stronger expression in many medial neurons, anterior neurons, and cells along the midline. The size of cells (white arrows) and their relative positions indicate that they are likely to be neurons. A–F, whole-mount preparations; G–L, filleted embryo preparations. Vertical white bars (G–L) indicate the ventral midline. Anterior is to the left (A–F) or to the top (G–L). Bars, 20 µm (A and D; B, C, E, and F; and G–I are at the same magnification).

The fru locus encodes a complex set of transcripts (Fig 1; RYNER et al. 1996 ; GOODWIN et al. 2000 ; USUI-AOKI et al. 2000 ). To better understand the embryonic pattern of fru transcripts, we performed in situ hybridization with a set of 5' end riboprobes to distinguish transcripts made from different fru promoters (see MATERIALS AND METHODS; RYNER et al. 1996 ; GOODWIN et al. 2000 ). We were unable to detect P1 or P2 transcripts at any embryonic stage using riboprobes specific to these transcripts (Table 2A). Thus, P1 transcripts, which encode the male-specific fru proteins, are not expressed in the embryo, a finding consistent with LEE et al. 2000 . Transcripts from P3 and P4 promoters were expressed during embryogenesis in a temporal and spatial pattern that mirrored that of fru transcripts detected by antisense-BTB or -Com riboprobes (Table 2A). In the developing CNS, both P3- and P4-specific riboprobes labeled mesectodermal and neuroectodermal cells followed by labeling of delaminating neuroblasts (Table 2A). At slightly later stages (st 9–11), medial neuroblasts continued to express P3 and P4 transcripts, but lateral neuroblasts no longer had detectable transcripts. In stages 7–12 embryos, the in situ hybridization signal for P4 transcripts was less intense, suggesting that the level of P4 transcripts was lower than that of P3 transcripts in these stages. However, P3 transcript levels became undetectable in all tissues after stage 12, whereas P4 transcripts were still detectable up to stage 16 (Table 2A). In summary, the in situ hybridization data show that both P3 and P4 fru transcripts are expressed in the developing CNS and, very likely, in the same cells. Furthermore, the higher level of fru transcripts detected with riboprobes to fru common sequences in stages 9–12 CNSs is likely due to the presence of both P3 and P4 transcripts whereas the lower level of fru transcripts in stages 12–16 CNSs reflects the presence of only P4 transcripts.

An additional complexity in fru transcripts reflects alternative splicing at the 3' end, which generates transcripts containing one of three different pairs of Zn-finger domains (Fig 1; RYNER et al. 1996 ; GOODWIN et al. 2000 ; USUI-AOKI et al. 2000 ). While it is known that P1 transcript isoforms are spliced to each of the three alternative 3' ends (GOODWIN et al. 2000 ), the full complexity of the 3' alternative splicing of transcripts produced from the P2, P3, and P4 promoters is not known. We examined embryos by in situ hybridization with three different 3' end riboprobes to detect transcripts having the A, B, or C 3' ends (MATERIALS AND METHODS; RYNER et al. 1996 ; GOODWIN et al. 2000 ). Overall, the temporal and spatial expression pattern of transcripts containing the A, B, and C 3' ends was consistent with the pattern found for fru transcripts labeled with antisense-BTB and antisense-Com riboprobes (Table 2A). In the developing CNS, transcripts containing the C 3' end appeared to be more abundant than those containing the A or B 3' ends (data not shown). In addition, some tissues, such as the tracheal placodes and amnioserosal cells, were labeled only when riboprobes to the C 3' end were used, suggesting that there is some tissue-specific regulation of the 3' alternative splicing of P3 and or P4 fru primary transcripts (Table 2A).

To confirm that we were detecting authentic fru transcripts, fru mutant embryos, fru^w24/fru^w24 and fru^sat15/fru^sat15, in which the coding regions of the fru gene are deleted, were labeled with the antisense-BTB, antisense-Com, and antisense-A 3' end riboprobes. At the beginning of embryogenesis, we detected transcripts in these null mutant embryos only up to stage 5; later stage embryos were not labeled (Fig 2C and Fig F). By contrast, fru transcripts in wild-type embryos are present up to stage 16 (Table 2A). From these results, we infer that fru transcripts in very early mutant and wild-type embryos are maternally derived, but that the transcripts found only in older wild-type embryos are zygotically generated. The presence of maternal and zygotic fru transcripts was consistent with the results of Fru protein immunohistochemistry (see below).

The temporal and spatial expression of Fru protein during embryogenesis:
To determine whether the fru transcripts present during embryogenesis are translated, embryos were labeled with antibodies directed against the common coding regions of Fru proteins (anti-Fru^COM and anti-Fru^BTB') and against the class A 3' end (anti-Fru^A') and the class C 3' end (anti-Fru^C'; see MATERIALS AND METHODS). We found that the localization of Fru proteins during embryogenesis is largely consistent with the spatial and temporal RNA distribution described above (Table 2A). Before neuroblast delamination, mesectodermal cells (st 9), but not neuroectodermal cells, were labeled with both anti-Fru^COM and anti-Fru^BTB' antibodies (Fig 2E; Table 2B). All delaminating neuroblasts and their progeny appear to express Fru proteins (st 9–10; Fig 2J). Ganglion mother cells (GMCs) located in the anterior regions of each segment label more strongly for Fru proteins than do those in the posterior regions (st 10; Fig 2K). At later stages (st 11–16), many cells in the CNS were strongly labeled (Fig 2L).

To determine whether Fru proteins were expressed in neurons and/or glia of stage 13–16 embryos, we carried out double- and triple-labeling experiments with anti-Elav antibodies to identify Fru-positive neurons and with anti-Repo antibodies to identify Fru-positive lateral glia (Fig 3A and Fig B). From a detailed comparison of double- (n = 5) and triple-labeled (n = 3) CNSs, it appeared that all Elav-positive cells in the CNS coexpress Fru proteins, even though there were variable levels of Fru protein expression in individual cells (Fig 3C). Likewise, all Repo-positive cells in the CNS were also Fru positive (n = 5, Fig 3C). In these experiments, some midline cells were only Fru positive (arrows, Fig 3C and Fig F). By their location, these cells are likely to be midline glia, which do not express Repo protein (HALTER et al. 1995 ). In summary, all neurons and glia appear to express Fru proteins.

View larger version (108K): In this window In a new window Download PPT slide   Figure 3. Coexpression of Fru protein with Elav, a neuronal marker, and Repo, a lateral glial marker, in the CNS and PNS. Filleted CNSs from wild-type embryos (st 16) were triple labeled with antibodies to Fru (red), Elav (green), and Repo (blue). A–C, D–F, and G–I are single confocal images in a dorsal CNS focal plane. The images that were merged in Adobe Photoshop are found in C, F, and I. J–L are composite merged images of the lateral body wall from embryos stained as above. (A) CNS cells labeled with anti-FruCOM antibody. The arrow points to a Fru-positive midline cell. (B) CNS cells labeled with anti-Elav and anti-Repo antibodies. (C) In this superimposed image of A and B, all neurons (yellow) and lateral glia (purple) colabel for Fru protein. The arrow points to the same Fru-positive midline cell as in A. (D) CNS cells labeled with anti-FruA' antibody. (E) CNS cells labeled with anti-Elav and anti-Repo antibodies. (F) In this superimposed image of D and E, most neurons (yellow) and glia (purple) colabel for FruA isoforms. The arrows point to Fru-positive midline cells shown in D. (G) CNS cells labeled with anti-FruC' antibody. (H) CNS cells labeled with anti-Elav and anti-Repo antibodies. Arrows point to two glial cells. (I) In this merged image of G and H, most neurons (yellow) but few glia (purple, arrows in H and I) colabel for FruC isoforms. (J) All peripheral sensory neurons (asterisk, yellow) are colabeled with anti-Elav and anti- FruBTB' but some glial cells are Fru positive (arrow, purple) and others are Fru negative (arrowhead, blue). The level of Fru expression is low in some sensory neurons and glial cells. (K) A few peripheral sensory neurons (arrowhead and asterisk, yellow) but many glial cells (arrow, purple) label for Fru isoforms that use the A 3' end. (L) Many peripheral sensory neurons (asterisk, yellow) but only a few glial cells (arrowhead, purple) label for Fru isoforms that use the C 3' end. Anterior is up for A–I and to the right for J–L. Bar, 20 µm.

We used antibodies specific to either the A or the C ZnF carboxy termini found in Fru isoforms to determine whether cells have Fru proteins with only one or both of these isoforms (see MATERIALS AND METHODS). The pattern of cells labeled with anti-Fru^A' is very similar to the spatial and temporal pattern of cells detected with anti-Fru^COM and anti-Fru^BTB' (Table 2B). In contrast, in young embryos (st 1–11) the anti-Fru^C' labeling pattern is similar to the pattern of cells stained with anti-Fru^COM and anti-Fru^BTB', whereas at later stages (st 12–16), fewer cells were labeled with anti-Fru^C' than with anti-Fru^COM and anti-Fru^BTB'. By double- and triple-labeling experiments in stage 16 embryos, all Elav-positive neurons, Repo-positive lateral glia, and putative midline glial cells were Fru^A positive (n = 4, Fig 3, D–F). However, all neurons, but only some Repo-positive glia, were Fru^C' positive (n = 5, Fig 3, G–I). These results suggest that all neurons and some lateral glia contain Fru isoforms having at least two different carboxy termini, but that some lateral glia and midline cells contain Fru proteins that may have only one type of carboxy terminus. Currently, no antibody specific to the B carboxy terminus is available.

In contrast to the findings from the in situ hybridization experiments, Fru proteins were detected in neurons and glial cells in the PNS by anti-Fru^BTB', anti-Fru^COM, anti-Fru^A', and anti- Fru^C' immunohistochemistry (Fig 3, J–L). In double- and triple-labeling experiments, some Fru-positive cells, labeled with anti-Fru^BTB' or anti-Fru^COM antibodies, were Elav positive and by location were judged to be mechanosensory and chordotonal organ neurons as well as the more internal multidendritic neurons. Another population of Fru-positive cells were Repo-positive and thus were identified as peripheral glial cells (Fig 3J; Table 2B). However, double- and triple-labeling experiments using anti-Fru^A' and anti-Fru^C' antibodies showed that only a subset of the Fru-positive sensory neurons and peripheral glia had Fru proteins with the A and C carboxy termini, from which we infer that some of these cells have Fru proteins with the B carboxy terminus (Fig 3K and Fig L; Table 2B).

Fru proteins are also expressed in other embryonic tissues. The posterior epidermis (st 13–16) and most body wall muscles (st 14–16, Table 2B) were Fru positive by anti- Fru^BTB' and anti-Fru^COM labeling. These tissues were labeled with anti-Fru^C' but not with anti-Fru^A' antibodies (Table 2B). These findings suggest that Fru protein isoforms with specific ZnF domains have a tissue-specific pattern in non-CNS tissues. The presence of Fru proteins in cell types that were not labeled in the in situ hybridization experiments, such as the PNS or muscle, may be due to differences in sensitivity of the molecular probes used and/or the relative levels of fru transcripts and proteins in these different cell types.

To be certain that authentic Fru proteins were being labeled in wild-type embryos, fru mutant embryos, fru^w24/fru^w24 and fru^sat15/fru^sat15, which are deleted for the fru gene, were labeled with anti-Fru^BTB', anti-Fru^COM, anti-Fru^A', and anti-Fru^C' antibody (Table 2B). At early embryonic stages (st 1–5), the pattern of Fru protein expression in null mutant embryos was comparable to that in wild-type embryos. At later stages (st 6–16), no Fru proteins were detected in null mutant embryos (Fig 2F), even though there was still robust labeling in wild-type embryos. These results indicate that the earliest transcripts present, which we infer to be maternal in origin, are translated in very early stage embryos and that Fru proteins in stage 6–16 embryos are largely or exclusively derived from zygotic fru transcripts.

Neurons in the embryonic CNS and PNS are labeled in fru P-element lines:
To determine whether P elements inserted within or near the fru locus behave as enhancer traps of the embryonic pattern of fru expression, we examined ß-galactosidase activity in fru³, fru⁴, and fru^sat mutant and AJ96w⁺ embryos (Fig 1; CASTRILLON et al. 1993 ; ITO et al. 1996 ; SPANA and DOE 1996 ). In fru³, fru⁴, and fru^sat mutations, a P element is inserted between the P2 and P3 promoters (Fig 1; CASTRILLON et al. 1993 ; ITO et al. 1996 ; RYNER et al. 1996 ; GOODWIN et al. 2000 ) and disrupts transcripts from the P1 and P2 fru promoters in pharate adult animals (GOODWIN et al. 2000 ).

In fru P-element mutant and control sibling embryos, cells in the CNS and PNS were labeled by anti-ß-galactosidase. Similar temporal and spatial labeling patterns were observed in embryos from the fru³ and fru⁴ lines. In embryos from these lines, anti-ß-galactosidase expression was detected in the mesectoderm (st 8), delaminating neuroblasts (st 9; Fig 4A and Fig B), and a variety of smaller cells, which appear by their location to be GMCs (st 10), neurons (st 11), and lateral glia and midline cells (st 12; Fig 4C). ß-Galactosidase labeling persisted in these CNS cells until stage 16. In the embryonic PNS, sensory organ precursors (st 10) followed by the external and chordotonal sensory neurons (st 12–16) were ß-galactosidase positive (Fig 4C). In addition, tracheal cells (st 11) and epidermal cells in the posterior part of each hemisegment (st 13–16) were labeled (Fig 4C). By contrast, in embryos from the fru^sat line, no ß-galactosidase-positive cells were found in the embryonic CNS, and only the chordotonal neurons in the PNS were labeled (data not shown). In summary, these results indicate that ß-galactosidase expression from the P elements inserted in fru³ and fru⁴ lines largely replicates fru's pattern in the embryonic CNS and PNS.

View larger version (114K): In this window In a new window Download PPT slide   Figure 4. ß-Galactosidase expression pattern in fru P-element mutant embryos. Whole-mount embryo (A) or filleted CNSs (B–D) from embryos labeled with anti-ß-galactosidase followed by AP histochemistry. (A) Whole-mount fru4 embryo (stage 9, ventral view). Cephalic (arrow) and ventral neuroblasts strongly express ß-galactosidase. (B) Filleted fru4 embryo (early stage 10, dorsal view). All ventral neuroblasts, including S1 neuroblasts (l, lateral column and m, medial column) and SII neuroblasts (i, intermediate column), are uniformly labeled. (C) Filleted fru4 embryo (stage 13, composite of dorsal to ventral views to show ventral nerve cord and lateral tissues). At later stages, the expression pattern in the ventral nerve cord shifts to a small number of midline cells (large arrows) and lateral cells, which are likely to be both glia (small arrows) and neurons based on the focal plane. Sensory neurons (white arrowheads), peripheral glia (small black arrowhead), and epidermis (white brackets) also strongly express ß-galactosidase. (D) Filleted AJ96w+embryo (stage 13). Two midline neurons, dMP2 and vMP2, are strongly labeled (arrows; SPANA and DOE 1996 ) as are sensory neurons (arrowhead). Vertical black bar (C and D) indicates the ventral midline of embryo. Anterior is up. Bars, 20 µm (C and D, same magnification).

We also examined the expression pattern of the enhancer trap line AJ96w⁺. Previous studies have shown that the MP2 neuroblast and its progeny, the dorsal and ventral midline precursor neurons (dMP2 and vMP2), express ß-galactosidase in the AJ96w⁺ line (SPANA and DOE 1996 ). The ventral CNS also showed very weak generalized neuronal expression from stage 14 to 16 and expression in a few lateral sensory neurons (Fig 4D). The generalized ventral nerve cord and sensory neuron labeling in embryos of the AJ96w⁺ line is similar to the pattern of expression in the fru³and fru⁴ lines, but these two fru lines did not show prominent MP2 or dMP2/vMP2 neuronal staining.

fru function is required for the formation of axonal tracts within the CNS:
Given the findings that the fru gene is widely expressed in the embryonic CNS, we expected that fru mutant embryos would have defects in CNS development and used two antibodies, anti-FasII and mAb BP102, to assay axonal projections within the CNS. FasII is a neural adhesion molecule expressed on the cell surface of axons forming specific longitudinal fascicles or tracts running throughout the entire ventral nerve cord and into the brain (Fig 5A; GRENNINGLOH et al. 1991 ; GOODMAN and DOE 1993 ; LIN et al. 1994 ; HIDALGO and BRAND 1997 ). In whole-mount or flat-dissected preparations of stage 16 and 17 wild-type embryos, three tracts, medial, intermediate, and lateral, are visible in the CNS.

View larger version (145K): In this window In a new window Download PPT slide   Figure 5. Axonal defects in the CNS of fru mutant embryos. Filleted CNSs from wild-type and fru mutant embryos (late stage 16) were stained for anti-FasII (A–C) and BP102 (D–F) using HRP histochemistry. (A) In a wild-type embryo, three bilaterally symmetric FasII longitudinal fascicles are visible. (B) In a fruw12/frusat15 embryo, all fascicles within a segment are disrupted. Brackets indicate the area shown at higher magnification in C. (C) Axons in all three fascicles have defasciculated. Axons in the medial MP1 fascicle extend toward the midline (arrowhead and arrow). (D) In a wild-type embryo, BP102-positive axonal processes in each segment form a bilaterally symmetric pair of longitudinal connectives along with the anterior and posterior commissures. (E) In a fruw12/frusat15 embryo, neither the longitudinal connectives (arrow) nor the commissures are uniform in size, suggesting that unequal numbers of axons are present. In some segments (arrow), the commissures are missing. (F) A higher power view of bracketed region in E to show the lack of axonal processes crossing the midline (arrowhead). Bars, 20 µm (A, B, D, and E, same magnification; C and F, same magnification).

FasII-positive tracts are abnormal in fru mutants lacking all or most fru transcripts. Between 12 and 25% of fru null mutant embryos (e.g., fru^sat15/fru^AJ96u3) had abnormal FasII-positive tracts (Fig 5B and Fig C; Table 3D). By comparison, <3% of wild-type embryos had any segments with disrupted FasII tracts (Table 3A). In all fru null mutant CNSs, FasII-positive axons no longer formed distinct tracts in one or more adjacent hemisegment, suggesting that these axons had defasciculated from other axons within the tracts. In some cases, axons that had defasciculated crossed and joined an adjacent fascicle or approached and crossed the midline (Fig 5C). In other cases, the left and right medial tracts appeared to merge along the midline.

To determine whether many or most axonal tracts were disrupted in fru mutants, we labeled the longitudinal connectives and commissures with the BP102 antibody (Fig 5D; SEEGER et al. 1993 ). Almost 20% of the fru null mutant embryos had defects in the pattern and distribution of BP102-positive axons in the connectives and commissures compared to only 1% of wild-type embryos (Table 3A and Table 3D). Most commonly, in these mutants, the commissures and connectives were not uniform but were either thicker, as though more axons were present, or thinner, as though fewer axons were present (Fig 5E and Fig F; Table 3C and Table 3D).

To demonstrate that these FasII and BP102 axonal defects depended on the loss of fru function, we used the fru^w12 allele, which has a chromosomal break within the fru locus, in combination with fru deletion mutations (Fig 1; Table 3C). We found that 15–23% of these mutant embryos had defects in their FasII and BP102 tracts; the frequency and the severity of the defects were similar to those found in fru null embryos (Fig 5B and Fig C). The fru^w12 allele is caused by a chromosomal inversion, and to rule out the possibility that the non-fru inversion breakpoint contributes to the FasII mutant phenotype, we examined fru^w12/Df(3L)XD198 mutants and found that the pattern of FasII fascicles was wild type (n = 10; data not shown; see MATERIALS AND METHODS). These results show that it is the loss of fru function that causes the axonal defects in FasII and BP102 tracts (Table 3C and Table 3D). The chromosomal break in the fru^w12 allele in the fru locus separates the P1, P2, and P3 promoters from the fru coding region but leaves the P4 fru transcription unit intact (ANAND et al. 2001 ). This result suggests that fru transcripts from the P1, P2, or P3 promoters are important for wild-type axonal pathfinding.

To further define which fru transcripts are required for the formation of FasII and BP102 tracts, we examined fru mutants in which P1 transcripts are affected or where P1 and P2 transcripts are eliminated (e.g., fru^4-40/fru^sat15; Table 3B; ANAND et al. 2001 ). These mutants had wild-type FasII and BP102 axonal tracts and produced P3 and P4 transcripts (Table 3B). The finding that P1 and P2 fru transcripts are not required for the formation of FasII and BP102 tracts is consistent with evidence by in situ hybridization that these transcripts are not present in embryos (Table 2A). By considering the different fru mutant genotypes examined, we infer that expression of P3 and, perhaps, P4 fru transcripts is sufficient for the development of wild-type FasII and BP102 tracts. The role of P4 transcripts in this process is inferred from data of mutants expressing the fru^w12 allele and it is possible that P4 transcripts, while present, are not expressed as in wild-type animals (ANAND et al. 2001 ).

fru function is required for wild-type orientation of pioneering axonal projections:
If the axonal pathfinding defects in fru mutants are due to the loss of fru function in neuronal precursors or in neurons themselves, then there are two likely explanations for the altered axonal trajectories in these mutants. One explanation is that neurons have not adopted, or only partially adopted, their wild-type identity in fru mutants and thus their axons fasciculate with different axonal partners as they grow in the CNS. An alternative explanation is that neurons in fru mutants adopt their wild-type identity but are unable to carry out their normal program of axonal pathfinding and differentiation. To distinguish between these two possibilities, we used neuronal and axonal markers to identify the earliest developmental abnormalities in fru mutants.

To assess whether fru played a role in neuroblast delamination and identity, we first labeled developing neuroblasts with antibodies to the Hunchback (Hb) protein, which labels all delaminating neuroblasts by early stage 9 in wild-type embryos (Fig 6A, n = 10). Fewer neuroblasts were labeled with anti-Hb antibody in early stage 9 fru^w12/fru^sat15 embryos than in wild type, but neuroblasts in all three rows did become Hb positive in late stage 9 embryos (Fig 6B and Fig C; n = 10 both stages). Thus, the final pattern of Hb expression was wild type in these fru mutant embryos but there was a slight temporal delay in either neuroblast delamination itself or the onset of Hb expression in delaminating neuroblasts.

View larger version (178K): In this window In a new window Download PPT slide   Figure 6. Neuroblast and neuronal identity and early axonal projections in the CNS of fru mutant embryos. Filleted CNSs from wild-type and fru mutant embryos labeled with anti-Hunchback (A–C), mab22C10 (D, E, G, and H), anti-Oddskipped (F), and anti-FasII (I–L) antibodies using AP (A–H) or HRP histochemistry (I and J). (A) Early stage 9 wild-type CNS. Full complement of Hb-positive NBs in medial (m), intermediate (i), and lateral (l) columns. Rows of En-positive cells (arrowheads) indicate the position of posterior compartments in A–C. (B) Early stage 9 fruw12/fruw24 CNS. Fewer Hb-positive neuroblasts (arrow) are found at this stage than in wild-type embryos. (C) Late stage 9 fruw12/fruw24 CNS. Hb-positive neuroblasts in all columns are present in a pattern similar to that in wild-type embryos. (D) Midstage 12 wild-type CNS. The mab22C10-positive vMP2 neuron (white arrow) projects its axon anteriorly (white arrowhead) and the dMP2 neuron (arrow) initially projects its axon posteriorly (arrowhead). (E) Midstage 12 fruw12/frusat15 CNS. vMP2 and dMP2 neurons often do not express mab22C10 as strongly as does wild type. In some segments, mab22C10-positive neurons were not in their typical anterior-posterior location (indicated by asterisk) and had delayed outgrowth of axonal processes (white arrowhead). The white arrow points to a cell with an abnormally oriented growth cone. (F) Midstage 12 fruw12/frusat15 CNS. Odd-skipped-positive vMP2 (white arrowhead) and dMP2 (white arrow) neurons are present in the fru mutant embryo. (G) Early stage 13 wild-type CNS. Initial axonal outgrowth forming the mab22C10 medial fascicle (white arrow). (H) Early stage 13 fruw12/frusat15 CNS. The medial fascicle (asterisks) is not yet formed. (I) Midstage 12 wild-type CNS. FasII-positive aCC neuron (white arrow) extends its axon peripherally (white arrowhead) and the FasII-positive pCC neuron (arrow) projects its axon anteriorly (arrowhead). (J) Midstage 12 fruw12/P14 CNS. In the more anterior hemisegment, the FasII-positive aCC neuron (white arrow) extends its axonal growth cones toward the periphery (white arrowhead) and the pCC neuron (arrow) sends its axon anteriorly (arrowhead). In the other hemisegment both aCC and pCC axons extend anteriorly (asterisk). (K) Early stage 13 wild-type CNS. The developing FasII medial fascicle (arrow) is well organized. All aCC neurons have extended their axons posteriorly (white arrowheads). (L) Early stage 13 fruw12/P14 CNS. The developing FasII medial fascicle (arrow) is indistinct, suggesting that the axons are not as fasciculated as those in wild type. In one hemisegment, the FasII-positive aCC axon (white arrow) extends posteriorly, not toward the periphery. Other aCC axons (white arrowhead) extended to the periphery. Anterior is to the top. The midline is to the right (I–L). Bars, 20 µm (A–C, same magnification; D–L, same magnification).

We next examined the development of neurons in fru mutants that pioneer the FasII fascicles to better assess whether there were changes in neuronal identity or early defects in axonal differentiation. In the wild-type CNS, pCC, vMP2, dMP2, and MP1 axons initiate the formation of the medial and intermediate FasII tracts within each segment (HIDALGO and BRAND 1997 ; HIDALGO and BOOTH 2000 ). The axonal process of the pioneer neurons vMP2 and pCC ascends while the processes of the MP1 and dMP2 axons fasciculate and extend posteriorly. These axons initially produce one fascicle at stage 13, which then splits into two fascicles, the pCC/vMP2 (medial) and dMP2/MP1 (intermediate) fascicles.

We therefore examined the development of aCC, pCC, vMP2, and dMP2 neurons in fru^sat15/fru^AJ96u3 and fru^w12/fru^AJ96u3 embryos. Along the midline, the MP2 precursor expresses the Odd-skipped protein and upon division the dMP2 daughter cell maintains Odd-skipped expression while the vMP2 daughter downregulates Odd-skipped expression (BROADUS et al. 1995 ; SPANA and DOE 1996 ). In fru mutant embryos all segments had the expected complement of Odd-skipped neurons along the midline (Fig 6F; fru^w12/fru^AJ96u3 and fru^sat15/fru^AJ96u3, n = 35 segments, n = 5 embryos of each genotype). The other pioneer neurons, aCC and pCC neurons, can be recognized by their location and FasII expression in stage ^<