A dual function of phyllopod in Drosophila external sensory organ development: cell fate specification of sensory organ precursor and its progeny

The Drosophila external sensory (es) organ is an excellent model system to study cell fate specification. Its development involves multiple steps of cell fate determination, which are governed by combined effects of intrinsic and extrinsic signals. Eventually, these signals are transduced to the nucleus and integrated to change the state of gene expression that promotes a particular fate (Jan and Jan, 1994). In the formation of an es organ, the first step is expression of proneural genes in a small group of ectodermal cells, known as ‘proneural clusters’. achaete (ac) and scute (sc) are the proneural genes that promote es organ formation, and both genes encode basic helix-loop-helix (bHLH) transcriptional activators (for reviews, see Ghysen and Dambly-Chaudiere, 1988; Campuzano and Modollel, 1992). Although each cell in the proneural cluster is competent to form a neural precursor, these proneural-cluster cells compete with each other so that a single cell is selected to develop into a sensory organ precursor (SOP). The remaining cells are prevented from adopting the SOP neural fate by the process of lateral inhibition that involves cell-cell interaction (for reviews, see Greenwald, 1998; Artavanis-Tsakonas et al., 1999) mediated by the receptor Notch (N) and its ligand Delta. Upon ligand binding, the membrane receptor N is processed, and the intracellular domain (N^ICD) (Schroeter et al., 1998) is released from the membrane to the nucleus (Lecourtois and Schweisguth, 1998; Struhl and Adachi, 1998). N^ICD forms a complex with the DNA-binding protein Suppressor of Hairless [Su(H)] to activate transcription of the Enhancer of split complex [E(spl)-C] (Bailey and Posakony, 1995; Lecourtois and Schweisguth, 1995). Members of the E(spl)-C down-regulate ac and sc expression in cooperation with the transcriptional repressor Groucho (Heitzler et al., 1996), thus leading to suppression of the SOP cell fate. In neurogenic mutants that disrupt this lateral inhibition process, supernumerary cells adopt SOP rather than alternative epidermal cell fate.

An adult es organ (bristle) consists of four different daughter cells (neuron, sheath cell, hair cell and socket cell) that are generated through asymmetric division of a SOP cell (Gho and Schweisguth, 1998; Gho et al., 1999; Reddy and Rodrigues, 1999). After the first asymmetric division, the SOP cell produces two secondary precursors, IIa and IIb (see schematic drawing in Fig. 7A). The IIb cell again divides asymmetrically to generate a precursor IIIb and a glial cell that migrates away soon after, and the IIa cell divides to generate a hair cell and a socket cell (the two outer support cells). Finally, the IIIb precursor divides to produce a neuron and a sheath cell. Proteins encoded by numb and N play a key role in the binary fate determination. Numb, a membrane-associated protein, localizes to a crescent during mitosis and segregates preferentially to one daughter cell after cell division (Rhyu et al., 1994; Knoblich et al., 1995). In numb loss-of-function mutants, cell fate transformation from IIb cells to IIa cells, hair cells to socket cells, and neurons to sheath cells occur (Uemura et al., 1989; Rhyu et al., 1994; Wang et al., 1997). By contrast, misexpression of numb results in cell fate transformation that is opposite to the phenotype observed in numb mutants (Rhyu et al., 1994). Biochemical and genetic analyses suggest that numb controls cell fate at least in part by antagonizing N activity (Guo et al., 1996). The asymmetrically segregated Numb protein prevents N activation in the daughter cell that inherits the Numb protein. Activation of N in the other daughter cell leads to activation of downstream signaling events that specify a cell fate distinct from that of its sibling. The zinc-finger protein Tramtrack (Ttk) is one of the nuclear targets to receive such N signaling (Guo et al., 1995; Guo et al., 1996). The ttk gene encodes two alternatively spliced forms of protein, Ttk69 and Ttk88 (Read and Manley, 1992), and both forms act as transcriptional repressors (Read et al., 1992; Xiong and Montell, 1993). The roles of ttk in cell fate determination have been closely examined in the developing eye and peripheral nervous system (PNS). In the eye, loss of ttk88 function results in formation of supernumerary R7 cells (Xiong and Montell, 1993; Li et al., 1997). In es organ development, ttk is expressed in the outer support cells and sheath cells, and loss of ttk function results in transformation from IIa to IIb cells and from sheath cells to neurons (Guo et al., 1995). Misexpression of ttk causes a phenotype opposite to that of ttk mutants in the es organ lineage (Guo et al., 1995), indicating that ttk acts as a binary cell fate switch in these asymmetric cell divisions. Genetic analyses show that ttk functions downstream of numb and N, and that Ttk protein levels are positively regulated by N (Guo et al., 1996)

phyllopod (phyl) was identified as being required for the cell fate determination of photoreceptor cells (Chang et al., 1995; Dickson et al., 1995). In the absence of phyl, R1, R6 and R7 cells are transformed into additional cone cells, and misexpression of phyl in cone cell precursors transforms these cells into additional R7 cells. During eye development, phyl is specifically expressed in R1, R6 and R7 cells, and its transcription is up-regulated by RAS1/MAPK signaling. Yeast two-hybrid and in vitro binding assays revealed that Phyl, Ttk and Seven in absentia (Sina), which is another protein required for fate determination of R7 cells (Carthew and Rubin, 1990), interact directly with each other (Li et al., 1997; Tang et al., 1997). During eye development, misexpression of phyl down-regulates the Ttk protein level in a sina-dependent manner (Tang et al., 1997). Cell culture experiments also showed that Ttk protein degradation is enhanced when both Phyl and Sina are present, and is dependent on the ubiquitin/proteasome pathway (Li et al., 1997; Tang et al., 1997). Together, these results suggest that Phyl and Sina function together to promote cell fate determination of photoreceptors by targeting Ttk for protein degradation.

In this report, we describe the roles of phyl in cell fate specification of SOP and SOP daughter cells in es organ development. We present our characterization of phyl loss-of-function and phyl misexpression phenotypes in both embryonic and adult es organ development. Also, we show that phyl interacts genetically with sina, and promotes es organ development by antagonizing ttk activity. phyl acts epistatically to N in cell fate specification of both SOP and its daughter cells. phyl is expressed in cells with neural potential (neuroblast, SOP and IIb cells), and its expression is negatively regulated by N signaling.

The following mutant alleles were used: phyl¹, phyl², phyl⁴ (Dickson et al., 1995), phyl²²⁴⁵ (Chang et al., 1995), sina², sina³ (Carthew and Rubin, 1990), ttk^lell (Xiong and Montell, 1993), ttk^osn (Guo et al., 1995), N^55e11 and E(spl)^{b32.2 gro+} (Heitzler et al., 1996). lacZ reporter genes, A101 (Usui and Kimura, 1993) and ase-lacZ (Jarman et al., 1993), were used for labeling the SOP cells; A1(2)29 (Bier et al., 1989) for outer support cells of es organs, and ac-lacZ (Gomez-Skarmeta et al., 1995) for the proneural clusters. The GAL4 lines used were: sca-GAL4 (Hinz et al., 1994), GAL4^109-68 (Frise et al., 1996), dpp-GAL4 (Masucci et al., 1990; Staehling-Hampton et al., 1994) and Eq-GAL4 (this study).

To produce phyl mutant clones, yw hs-FLP; FRT^42d phyl²²⁴⁵/FRT^42d y+, FRT^42d pwn phyl²/FRT^42d; hs-FLP Sb/+ and yw UAS-FLP; sca-GAL4 FRT^42d phyl²²⁴⁵/FRT^42d y+ flies were generated. To generate phyl mutant clones in N^ts flies, females of N^ts omb^p1/+; FRT^42d p[conD]/+ were crossed with males of FRT^42d pwn phyl²; hs-FLP Sb/SM5-TM6B. Larvae at 24-48 hours after egg laying were heat treated at 37°C for 15 minutes twice to induce the mutant clones. Supernumerary microchaetes were induced by incubating the N^ts pupae (0-6 hours after puparium formation; APF) at 30°C for 6 hours.

For misexpression of phyl, five independent UAS-phyl transgenic lines were generated. All lines gave similar phenotypes. UAS-phyl 1b and 2a gave stronger phenotypes and were used for further experiments. UAS-N^ACT (Doherty et al., 1996) and UAS-ttk69 (Lai and Li, 1999) were also used in this study.

For heat shock treatment of N^ts embryos, embryos collected between 10-15 hours at 17°C were incubated at 32°C for another 5 hours to eliminate the N activity, and were then placed at 25°C for another 2.5 hours prior to fixation.

Embryos were fixed and stained as described previously (Chien et al., 1998). The primary antibodies used were mouse anti-Elav antibody (1:200), rabbit anti-Prospero antibody (1:1000) (Vaessin et al., 1991), rat anti-Cut antibody (1:2000) (Blochlinger et al., 1990), rat anti-Su(H) antibody (1:1000) (Gho et al., 1996), and mouse and rabbit anti-β-galactosidase antibodies (1:250 and 1:1000, respectively). For antibody staining of pupal nota, nota were dissected in 1× PBS and fixed in 4% paraformaldehyde. After washing with 1× PBS, nota were then incubated with primary antibodies: mouse anti-Elav antibody (1:50), rabbit anti-Prospero antibody (1:200) and rat anti-Cut antibody (1:1000), followed by Cy3, Cy5, Alexa-488, or FITC-conjugated secondary antibodies.

Imaginal disks and pupal nota were dissected in 1× PBS, and fixed in 0.3% glutaraldehyde solution (in 1× PBS) for 10 minutes. Disks and nota were washed in PBS for 15-30 minutes before staining in 0.3% X-gal in Fe/NaP solution.

The procedure used for in situ hybridization has been described by Tautz and Pfeifle (Tautz and Pfeifle, 1989). Digoxigenin-labelled RNA probes of phyl and lacZ were prepared as described previously (Lehmann and Tautz, 1994).

Previous reports (Chang et al., 1995; Dickson et al., 1995) have described phyl mutations causing defective sensory bristles in adult flies. However, no detailed analyses have been performed. We examined the phenotypes in adults and embryos homozygous or transheterozygous for several loss-of-function phyl alleles, phyl¹, phyl², phyl⁴ (Dickson et al., 1995) and phyl²²⁴⁵ (Chang et al., 1995). According to their mutant phenotypes in adult bristles and the embryonic PNS (see below), phyl¹, phyl² have been classified as strong loss-of-function alleles, phyl²²⁴⁵ as a slightly weaker allele, and phyl⁴ as a weak hypomorphic allele. phyl¹, phyl², and phyl²²⁴⁵ alleles were found to be lethal as homozygotes or in heteroallelic combinations. However, the phyl¹/phyl⁴, phyl²/phyl⁴ and phyl²²⁴⁵/phyl⁴ animals were semiviable, and the surviving adults exhibited two distinct classes of defects on sensory bristles. The first class of defects was missing bristles. For example, approximately 78% of the notal and 87% of the abdominal microchaetes were missing in phyl¹/phyl⁴ flies (Fig. 1B,D; Table 1). The second class of defect associated with phyl mutations was abnormal configuration of the bristles (arrows in Fig. 1B,D; Table 1). The most prominent phenotype of this class on the notum was duplicated bristles (e.g., 45% in phyl¹/phyl⁴ animals), with two shafts emerging from either two sockets or a fused socket. Other phenotypes included abnormal bristles with 1 shaft surrounded by 2-3 sockets, and 2- or 4-socket clusters without shafts. In a weaker phyl mutant (phyl²²⁴⁵/phyl⁴), the missing bristle phenotype was less prominent, but the number of abnormal bristles was increased (data not shown). Similar phenotypes (missing and abnormal bristles) were also observed in phyl²²⁴⁵ and phyl² mutant clones (data not shown and Fig. 5D, respectively).

The missing bristle phenotype in phyl mutants might result from failure of SOP formation. During SOP formation, the first step is the expression of proneural genes in the proneural clusters. Thus, we examined achaete-lacZ (ac-lacZ) expression in wild-type and phyl mutant nota at 14-18 hours APF. In the wild-type notum, ac-lacZ was expressed in a regular array of cells, which were the precursors of adult microchaetes (Fig. 1E). In the phyl mutant, the ac-lacZ transgene was also expressed in a pattern analogous to that of the wild type (Fig. 1F). These data indicate that the missing bristle phenotype caused by phyl mutations does not result from a failure of proneural gene expression.

Closer examination of phyl mutants indicated that at 14-18 hours APF, ac-lacZ expression remained in clusters of cells (Fig. 1F, inset), while in the wild type animal at the same stage, lacZ expression was restricted to single cells (Fig. 1E, inset). This observation suggests that phyl may function at a later stage than proneural genes to promote SOP formation. We tested this possibility by examining the SOP markers. The enhancer trap line A101 drives lacZ expression in SOP cells and is an early marker for SOP cell fate (Huang et al., 1991) (Fig. 1G). In phyl mutants, most of the A101 expression was eliminated (Fig. 1H). We also examined the expression of another SOP marker, asense-lacZ (ase-lacZ). asense encodes a bHLH protein that is specifically expressed in SOP cells (Brand et al., 1993), and the ase-lacZ transgene was detected at a later stage than A101 (see Fig. 1I and Figure legends). In phyl mutants, most of the X-gal staining was absent (Fig. 1J). We observed similar results for macrochaetal SOP cells in the wing disks in which staining for A101 disappeared in phyl mutants (data not shown). Anti-Cut antibody can recognize all daughter cells of the es organ (Blochlinger et al., 1990). Consistently, only a few Cut-positive clusters were seen in pupal nota of phyl mutants (data not shown). These results indicate that phyl is essential for the fate specification of SOP cells.

In phyl mutants, some SOP cells formed and developed into bristles with abnormal compositions (Fig. 1B,D; Table 1), which was likely due to fate transformation of the sibling cells within the es organ lineage. To determine whether phyl is required for the binary cell fate decisions in the es organ lineage, we examined the composition of es organs in phyl mutants during embryonic development. In phyl mutant embryos, the numbers of the sensory neurons and the sheath cells, recognized by anti-Elav and anti-Prospero antibodies, respectively, were significantly reduced (compare Fig. 2B with Fig. 2A). In strong phyl mutants (phyl¹/phyl²), 75% of the neurons and 50% of the sheath cells were missing. We focused subsequent studies on two es organs in the dorsal-most (dm) region of abdominal hemisegments (Fig. 2C,E) since they are well separated from the rest of the sensory organs, thus making it easier to identify their constituent cells. In strong loss-of-function phyl²²⁴⁵/phyl²²⁴⁵ and phyl¹/phyl² mutant embryos, more than 80% of the neurons and sheath cells in the dm region were missing (Fig. 2D). However, anti-Cut antibody staining revealed that only 12% of the es organs at the dm region were absent (Cut-negative) (Fig. 2F), which might have resulted from failure in SOP formation. These results suggest that in phyl mutant embryos, most of the SOP cells of the dm es organs form and divide to produce four Cut-positive cells, and that the large number of missing neurons and sheath cells is not due to failure of SOP formation.

We next examined whether the absence of neurons and sheath cells in phyl mutants is the result of changes in cell identity. An enhancer trap line, A1(2)29, was used to label the outer support cells (hair cells and socket cells). In wild-type embryos, the two es organs in the dm region contained four outer support cells (Fig. 2C). In phyl mutants, most hemisegments contained more than four outer support cells; as shown in Fig. 2D, there were six outer support cells accompanied by one neuron and one sheath cell in the left hemisegment, and eight outer support cells in the right hemisegment. These results suggest that the neurons and the sheath cells were transformed into outer support cells in phyl mutants. We further examined whether phyl was required for binary cell fate specification between neurons and sheath cells, and between hair cells and socket cells. For the dm es organs, cell fate transformation from hair cells to socket cells [recognized by the anti-Su(H) antibody] occurred in less than 5%, and transformation from neurons to sheath cells was rarely seen (data not shown).

In phyl mutant adults, most SOP cells of notal bristles were missing, while in phyl mutant embryos, the majority of SOP cells formed in the dm es organs. One likely reason for this difference is the masking of the embryonic SOP phenotype by maternally deposited phyl transcripts. When both maternal and zygotic phyl mRNA were removed, however, the defects were similar to the zygotic mutant phenotypes; 75% of the neurons and 60% of the sheath cells in the abdominal hemisegments, and 10% of the dm es organs were missing (data not shown). Furthermore, in phyl¹/phyl² mutants, the staining pattern with anti-Cut antibody showed that, in addition to the dm es organs, most es organs of the embryonic PNS formed (data not shown). These results suggest that phyl is partly required for the fate specification of SOP cells in embryonic es organ development.

In phyl mutant adults, the phenotype of missing bristles is due to failure in SOP formation. To test whether misexpression of phyl can promote SOP formation, we used Eq-GAL4 to drive phyl expression in the notum. Eq-GAL4 was expressed in the anterior of the presumptive notum in the wing disk (Fig. 3A), and of the pupal notum (6-8 hours APF) with stronger expression in the anterior midline region (indicated by arrows). When the phyl expression was driven by Eq-GAL4, we observed ectopic bristles in the notum (Fig. 3D). In general, there was a 30%-70% increase in the number of notal microchaetes, with a highest density in the midline region (Fig. 3F). When dpp-GAL4 was used to misexpress phyl, ectopic bristles were also observed on the scutellum (Fig. 3H) and the third wing vein (data not shown). These ectopic bristles were derived from ectopic SOP cells, as was evident from the appearance of ectopic β-gal-positive cells in the A101 scutellum (Fig. 3J, arrows). These data indicate that misexpression of phyl can promote ectopic SOP formation.

To determine whether phyl indeed functions as a genetic switch for the binary cell fate specification of SOP daughters, we examined the phenotypes of phyl misexpression in sensory organ lineage by two GAL4 lines, GAL4^109-68 (Frise et al., 1996) and sca-GAL4 (Hinz et al., 1994). When phyl was misexpressed by sca-GAL4 in embryonic PNS, supernumerary sensory neurons were generated (Fig. 3L). This increase in the number of neurons (approximately 50% increase) was due to fate transformation of outer support cells. As shown in Fig. 3M, in the dm region, two out of the three hemisegments had four neurons and four sheath cells, and the third hemisegment had 5 neurons and three sheath cells. Since only eight Cut-positive cells were observed in each dm region (Fig. 3N), these results indicate that the extra neurons and sheath cells are derived from transformation of IIa cells to IIb cells in both es organs.

A similar phenotype of cell fate transformation was also observed in adult bristles. As shown in Fig. 3O, the outer hair and socket structures of microchaetes were all missing when phyl was misexpressed by GAL4^109-68. Pupal nota stained with anti-Elav and anti-Prospero antibodies showed clusters of two neurons and two sheath cells (Fig. 3Q), in contrast to wild-type es organs with clusters of one neuron and one sheath cell (Fig. 3P). Clusters of 3-4 neurons with 0-2 sheath cells were occasionally observed (data not shown). Therefore, misexpression of phyl in both embryonic and pupal es organ lineages most often resulted in transformation of outer support cells (hair cells and socket cells) to internal cells (neurons and sheath cells). Together with the phyl loss-of-function mutant phenotypes, these data indicate that phyl functions as a binary cell fate determinant in the formation of IIa and IIb cells in es organ development.

During R7 photoreceptor differentiation, Sina forms a complex with Phyl to promote R7 cell fate by targeting the transcriptional repressor Ttk for degradation (Li et al., 1997; Tang et al., 1997). We found that sina is also involved in es organ development. Similar to phyl mutants, two distinct bristle phenotypes were observed in sina loss-of-function mutants (Fig. 4B,H and Table 1). In sina²/sina³ adults, 41% of the notal and 62% of the abdominal microchaetes were missing. Also, some of the remaining microchaetes showed abnormal phenotypes such as duplicated bristles (17% in the notum and 65% in the abdomen). To determine whether formation of SOP cells is affected in sina mutants, we examined the expression pattern of A101 and ase-lacZ in sina²/sina³ pupae. Compared to wild-type pattern (see Fig. 1G), in sina²/sina³ mutants, some A101-positive cells were missing near the anterior midline region (arrows in Fig. 4M) where the microchaetes were affected most in mutant adults (Fig. 4B). This result demonstrates that sina is required for the formation of some microchaetal SOP cells. To our surprise, ase-lacZ expression in sina mutants was mostly eliminated (Fig. 4N), despite the fact that only a fraction of bristles in sina²/sina³ adults are missing. Since ase is not essential for bristle formation (Jarman et al., 1993), the missing ase expression does not necessarily correlate with the missing bristles. Nevertheless, these data indicate that sina is important for ase expression in most SOP cells.

We next examined the functions of sina in embryonic es organ development. In sina²/sina³ mutants, embryonic development (including the development of es organs) was normal (data not shown), and the mutants survived to adulthood. However, when both the maternal and zygotic sina transcripts were removed, embryonic development was affected to different degrees. While some embryos showed obvious defects in their overall morphology, others exhibited no defects. In embryos with normal morphology but severe defects in the PNS, approximately 75% of the neurons and 60% of the sheath cells in the abdominal hemisegments were missing (Fig. 4O). In the dm region, 90% of the es organs were still present (Fig. 4P), suggesting that sina was only partly required for the formation of embryonic dm SOP cells, and the reduction in the number of neurons and sheath cells might result from cell fate transformation. Indeed, we found an increase in the number of outer support cells in the A1(2)29 enhancer trap line (data not shown) in these sina mutant embryos. These data indicate that sina is required for the cell fate determination of IIb cells in the embryonic es organ lineage.

Because phyl and sina mutations cause similar phenotypes, and ttk is also essential for the binary cell fate decision of SOP progeny, we explored the possibility that phyl, sina and ttk function in the same genetic pathway in es organ development. We first examined their genetic interactions in adult bristle development. When one copy of phyl²²⁴⁵ or phyl² alleles was introduced into the sina²/sina³ mutants, the number of microchaetes was reduced on both nota and abdomens. Also, the percentage of abnormal notal microchaetes was increased more than twofold (Fig. 4C,I; Table 1), indicating that phyl mutations dominantly enhanced the sina mutant phenotypes. We then tested whether ttk mutations suppress sina and phyl mutant phenotypes. We used two ttk mutant alleles: ttk^lell, a null allele (Xiong and Montell, 1993), and ttk^osn, in which the 69 kDa protein expression is absent and the level of the 88 kDa Ttk protein is reduced (Guo et al., 1995). When one copy of ttk^lell or ttk^osn was introduced into phyl²/phyl⁴ mutants, the numbers of both notal and abdominal bristles were significantly increased, and the percentage of abnormal notal bristles was largely reduced (Fig. 4F,L; Table 1). A similar suppression effect was also observed when one copy of the ttk^osn allele was introduced into sina²/sina³ mutants (Fig. 4D,J; Table 1). Therefore, our genetic analyses are most consistent with the notion that phyl, sina and ttk function in the same genetic pathway, and phyl and sina promote bristle development by antagonizing ttk activity. One exception is that removing one copy of ttk enhanced the abnormal bristle phenotype in phyl mutant abdomen (Fig. 4L and Table 1).

In embryonic PNS, phyl also showed strong genetic interaction with ttk. In phyl¹/phyl² mutant embryos, 75% of the neurons in abdominal hemisegment and all of the dm neurons were missing (see Fig. 2B). However, in phyl¹/phyl²; ttk^osn/+ embryos, only 10% of the PNS neurons were absent and each dm region contained at least one neuron (Fig. 4Q). This result shows that the ttk mutation strongly suppresses the phyl mutant phenotypes in embryonic PNS, and indicates that phyl promotes cell fate specification of embryonic PNS neurons by antagonizing ttk activity.

Our genetic analyses suggest that phyl and ttk function in the same genetic pathway in es organ development. In embryonic development, lack of ttk activity causes fate transformation in sensory organ lineages that leads to overproduction of sensory neurons (Fig. 5H). This phenotype is in contrast to that observed in phyl mutant embryos (see Fig. 2B). In embryos lacking both phyl and ttk activity, we observed the formation of supernumerary sensory neurons (Fig. 5I) that was indistinguishable from that of the ttk mutant embryos. This result suggests that ttk functions epistatically to phyl.

In adults, most microchaetes were missing when ttk was misexpressed by Eq-GAL4 (Fig. 5A). The lack of microchaetes was the result of failure of SOP formation as most A101-positive cells disappeared (Fig. 5C). When phyl and ttk were coexpressed by Eq-GAL4, most notal microchaetes were missing (Fig. 5B), suggesting that misexpression of ttk can suppress the ectopic bristle phenotype caused by misexpression of phyl. Therefore, these results are consistent with the model that ttk acts downstream of phyl in the formation of SOP cells.

In es organ development, N is required to single out the SOP cells and to specify the cell fate of hair cells and socket cells. In N mutants, many supernumerary SOP cells are formed and hair cells and socket cells are transformed into neurons (Hartenstein and Posakony, 1990). Since phyl mutant phenotypes are opposite to those of N mutants, and both genes acts upstream of ttk in es organ development (Fig. 5 and Guo et al., 1996), we were interested to examine their epistatic relationship. In the strong loss-of-function phyl² mutant clones, most bristles were missing and the few remaining ones had a four-socket phenotype (Fig. 5D). In contrast, when the activity of N was eliminated during the stage of SOP emergence (6-12 hours APF) by using the N^ts allele, the number of microchaetes was significantly increased (Fig. 5E). When a phyl² mutant clone was generated in the N^ts fly with N activity inactivated, most of the bristles were lost within the clone (Fig. 5F). This result shows that phyl is epistatic to N in SOP cell fate specification. We also tested whether phyl acts epistatically to N in cell fate specification of SOP progeny in embryos. In the embryos from the cross of N^ts/N^ts; phyl²/+ females with N^ts/Y; phyl¹/+ males, a quarter of them were N^ts and phyl double mutants and the rest were N^ts mutants. When these embryos were incubated at the permissive temperature (17°C), 24% of them (n=40) showed a strong reduction in the number of sensory neurons (<20 neurons per abdominal hemisegement; Fig. 5J). The rest of the embryos (76%, n=127) had a normal number of neurons (42 in each abdominal hemisegment). This result indicates that about one quarter of the embryos in this cross were indeed phyl mutants. To inactivate N activity during embryonic development, the embryos were incubated at the non-permissive temperature (32°C) when SOP cells and their progeny were undergoing asymmetric divisions (Guo et al., 1996). We found that 21% of the embryos (n=39) had fewer than 20 neurons in each abdominal hemisegment (Fig. 5L), and a small fraction of embryos (2.6%, n=5) exhibited an intermediate phenotype with 20-40 neurons (data not shown). Overproduction of neurons was seen in 76% of the embryos (n=143; Fig. 5K). In a control experiment, we confirmed that this phenotype of neuron overproduction was indistinguishable from that of N^ts embryos at 32°C (data not shown). Therefore, this data showed that phyl acts epistatically to N in cell fate specification of SOP daughter cells in embryos.

The distribution of phyl mRNA was examined by whole-mount in situ hybridization. Although weak, the expression of phyl in wing imaginal disks was detected in the SOP cells of wing margin bristles (Fig. 6A,B), notal macrochaetes (Fig. 6A,C), and other sensory organs (Fig. 6A). In leg disks, phyl mRNA was also detected in the precursors of the femoral chordotonal organs, as well as in external sensory SOP cells (data not shown). During embryogenesis, the maternally contributed phyl transcripts were ubiquitously present before cellularization (data not shown). During stages 9-11, phyl was expressed in neuroblasts (Fig. 6E) and the SOP cells (Fig. 6F). From stage 12 onward, phyl transcripts diminished gradually, but remained detectable in a subset of PNS cells at stages 12-14 (Fig. 6G). These cells were probably the IIb cells and their progeny (neurons and/or sheath cells), since at the same stage, phyl mRNA was no longer detected in numb mutants (Fig. 6H), in which the IIb cells were transformed to the IIa cells (Uemura et al., 1989).

Our genetic analyses suggest that phyl is acting downstream of N in es organ development. We were interested in examining whether N signaling regulates phyl expression. In wild type, phyl was expressed in SOP cells in both wing disks and embryos. In N^ts mutants at restrictive temperature, phyl mRNA was detected strongly in clusters of cells in wing disks (Fig. 6D). phyl was also expressed in many more cells in N (Fig. 6K) and E(spl)-C (Fig. 6L) mutant embryos at stage 11, demonstrating that N signaling negatively regulates phyl expression. Consistently, when the constitutively active form of N (N^ACT) in which the extracellular domain is deleted (Doherty et al., 1996) was expressed in the embryonic sensory organ lineage by sca-GAL4, the levels of phyl mRNA at stage 12 were also strongly reduced (Fig. 6I).

To further confirm that N negatively regulates phyl activity by repressing its expression, we tested whether misexpression of phyl rescues the defects caused by misexpression of N^ACT. phyl is essential for fate specification of IIb cells in the embryonic sensory organ lineage. When UAS-phyl and UAS-N^ACT were coexpressed in embryos by sca-GAL4, the total numbers of PNS neurons and sheath cells (both are progeny of IIb cells) on average were double that in sca-GAL4/+; UAS-N^ACT/+ embryos (Fig. 6M,N). This result shows that misexpression of phyl rescues the defects caused by N^ACT in the embryonic sensory organ development.

We found that phyl is essential for two steps in es organ development: in phyl mutants, SOP cells of the adult es organs fail to form (Fig. 7B), and IIb cells are transformed into IIa cells in both adult and embryonic es organs (Fig. 7C). Also, our genetic analyses are consistent with the model in which Phyl, in collaborating with Sina, promotes fate specification of SOP and IIb cells by degrading Ttk protein (Fig. 7D,E). Furthermore, we showed that phyl is expressed in SOP and IIb cells, and its mRNA level is negatively regulated by N signaling (Fig. 7D,E).

Our genetic analyses show that phyl functions together with sina to promote SOP formation by antagonizing ttk activity. These results suggest that degradation of the Ttk protein is a major function of Phyl in the cell fate specification of SOP cells. Consistent with this idea, we found that misexpression of ttk can inhibit the formation of SOP cells and suppress the ectopic bristle phenotype caused by misexpression of phyl. Several lines of evidence also indicate that Ttk functions as a repressor to inhibit SOP cell fate. (1) Ttk is expressed ubiquitously in the pupal notum except in SOP cells (Ramaekers et al., 1997). (2) In embryos, overexpression of ttk inhibits the formation of es organs (Guo et al., 1995). (3) Injection of ttk dsRNA results in extensive increase of neurons in embryonic PNS, a phenotype observed in neurogenic mutants (Kennerdell and Carthew, 1998). All of these results suggest that ttk might play a negative role in the fate specification of SOP cells, and phyl promotes SOP fate specification by degrading Ttk.

phyl is essential for formation of pupal SOP cells, but is partly required for embryonic SOP cells. In senseless mutants, the larval SOP cells fail to form, but the embryonic SOP cells form and divide to generate daughter cells that fail to differentiate (Nolo et al., 2000). These results suggest that there are some distinctions between the SOP cell fate specification of embryos and larvae.

Our studies showed that phyl is required for IIb cell fate specification. In phyl mutants, more than half of the adult notal microchaetes and more than 80% of the embryonic dm es organs exhibited IIb to IIa cell fate transformation. Cell fate transformation from hair cells to socket cells also occurs in adult bristle (four-socket phenotype seen in phyl² clone), but much less frequent in embryonic dm es organs (less than 5%). Transformation from neurons to sheath cells in es organs was rarely seen in phyl embryos. These results suggest that unlike numb and N, phyl is mainly required for the binary cell fate determination between IIa and IIb cells in both adult and embryonic es organs. In support of this conclusion, misexpression of phyl in both adult and embryonic sensory organ lineages most often resulted in a two-neuron/two-sheath cell phenotype.

In phyl¹/phyl² mutant embryos, most of the neurons and sheath cells in chordotonal organs were also lost (Fig. 2B, lateral and ventral region), indicating that phyl is also required for the formation of neurons and sheath cells in chordotonal organ lineage. In weaker mutant embryos (phyl²²⁴⁵/phyl²²⁴⁵), we often observed more sheath cells than neurons in the regions where chordotonal organs lch5 and vchA and B are located. Therefore, phyl may be required for cell fate determination between neurons and sheath cells in the development of chordotonal organs.

During lateral inhibition, the N pathway is essential to single out the SOP cells. Our in situ analyses indicate that phyl expression is negatively regulated by N. Also, the supernumerary bristles in N mutants were suppressed by phyl mutation. These results strongly suggest that down-regulation of phyl expression is a major function of N signaling to suppress SOP cell fate.

In the SOP lineage, N also plays an important role in the cell fate specification of SOP progeny. Several components of the N pathway, such as Delta, Serrate (Zeng et al., 1998), Su(H) (Schweisguth and Posakony, 1994), Hairless (Bang and Posakony, 1992), and proteins of the Bearded (Leviten and Posakony, 1996; Lai et al., 2000) and E(spl) (Tata and Hartley, 1995) families, have been shown to be involved in the cell fate specification of all or subsets of progeny. We found that phyl acts epistatically to N in the cell fate specification of SOP daughter cells and is expressed in IIb cells. Also, misexpression of phyl rescues the defects caused by N^ACT, indicating that N regulates the phyl activity in sensory organ lineage at the transcriptional level.

Our genetic analyses of phyl, sina and ttk are mostly consistent with the model that Phyl functions together with Sina to promote es organ development by degrading Ttk. In embryos, strong defects are only detected when both maternal and zygotic sina transcripts are removed, suggesting that maternally contributed sina transcript play an essential role in the development of embryonic es organs. Consistently, no genetic interaction between zygotic sina and phyl, and between zygotic sina and ttk was detected in embryonic PNS development (data not shown). In adults, the bristle phenotypes in sina mutants are weaker than in phyl mutants. One possible reason is that the perdurance of sina gene products from maternal transcripts might supply activity for some adult bristles to develop normally. Another possibility is that phyl is able to down-regulate ttk activity in a sina-independent manner. In the Drosophila genome, a sequence (CG 13030) is located next to sina in the genome and encodes a putative protein with 50% identity and 70% similarity with Sina. It might be possible that sina functions redundantly with this gene in bristle development.

Our studies of phyl/sina/ttk in es organ development and previous studies in photoreceptor differentiation indicate that the Drosophila eye and es organs depend on the same protein complex to specify their cell fate. In both cases, phyl mutations transform neural cells to non-neural cells. Both studies also show that phyl expression is tightly regulated by the upstream signaling pathways. The expression of phyl is activated by the Ras pathway in photoreceptor cells. In SOP cells, the transcription of phyl is likely activated by the proneural genes ac and sc, and is repressed by N signaling. Interestingly, it has been shown that the Egrf/Ras/Raf pathway acts antagonistically with the N pathway in SOP formation of adult macrochaetes (Culi et al., 2001) and chordotonal organs (zur Lage and Jarman, 1999). Whether these two pathways converge on phyl expression to regulate sensory organ formation remains to be examined.

We thank E. Hafen, P. Heitzler, J.-c. Hsu, Y. N. Jan, T. Laverty, V. Rodrigues, F. Schweisguth, Y. H. Sun, Bloomington Stock Center and the Developmental Studies Hybridoma Bank (University of Iowa) for antibodies and flies. We thank E. Hafen for phyl cDNA. We also thank Y.-H. Chou, M.-C. Li and J.-M. Bai for their help with whole-mount in situ hybridization and A.-C. Tien for his help with microinjection of the UAS-phyl construct. We are grateful to J.-c. Hsu, G.-j. Liaw and Y. H. Sun for critical reading of the manuscript and D. Platt for English editing. This work was supported by grants from Academia Sinica and the National Science Council of Taiwan to C.-T. Chien (NSC 89-2311-B-001-225). H. Pi is supported by a postdoctoral fellowship from Academia Sinica, Taiwan.