Homeotic genes have specific functional roles in the establishment of the Drosophila embryonic peripheral nervous system

The Drosophila embryonic PNS develops in a stereotypical manner and contains a wide variety of sensory organs. A typical sensory organ consists of a neuron(s) and three support cells which will form the sensory structure innervated by the neuron (Hertweck, 1931; McIver, 1985; Zacharuk, 1985; Hartenstein, 1988). The entire embryonic PNS has been characterized in such detail that the positions of each sensory neuron and the associated sensory structure can be identified in both the thoracic and abdominal segments (Campos-Ortega and Hartenstein, 1985; Dambly-Chaudière and Ghysen, 1986; Bodmer and Jan, 1987; Hartenstein, 1988). There are two types of sensory organs in the embryonic PNS: external sensory (es) organs which are believed to sense mechanical and chemical stimuli and chordotonal organs (ch), proprioceptors that are proposed to sense stretch (McIver, 1985; Zacharuk, 1985; Dambly-Chaudière and Ghysen, 1986; Bodmer et al., 1987).

The es organs are composed of at least one neuron which has a single dendrite and three support cells. The support cells are the thecogen, trichogen and tormogen which respectively form the sheath around the dendrite, the sensory process (bristle), and the circular socket of the sensory organ. The tormogen and trichogen cells are located in the epidermis, while the neuron(s) and thecogen are located beneath the epidermis. A scolopidium, the basic unit of ch organs, consists of four cells: a neuron with a single dendrite, the scolopale cell, cap cell, and ligament cell. The scolopale cell forms a sheath around the dendrite, while the cap cell and ligament cell mediate the attachment of the ch organ to the body wall. The prothoracic lateral triscolopidial (lch3) and abdominal pentascolopidial (lch5) organs have an additional pair of attachment cells at the apical end of the structure (Matthews et al., 1990; Ghysen and O’Kane, 1989; J. Heuer, unpublished observations, Fig. 1). All cells of ch organs are located beneath the epidermis.

It has been proposed that all cells comprising the es and ch sensory organs are lineally related (Bodmer et al., 1989) and arise through divisions of ectodermal precursor cells close to their final position (Bate, 1978; Bodmer et al., 1989). Despite their basic similarity, there are several characteristics which allow one to distinguish between es and ch organs including morphology, antigenic properties, division patterns and their final position with respect to the cuticle (Bodmer et al., 1987; Bodmer et al., 1989).

There are five distinct spatial patterns of sensory organs in the trunk of the segmented Drosophila embryo. Unique patterns are present in: (1) the prothoracic segment (Tl), (2) the meso- and metathoracic segments (T), (3) the first through the seventh abdominal segments (A), (4) the eighth abdominal segment (A8) and (5) the ninth abdominal segment (A9) (Campos-Ortega and Hartenstein, 1985; Dambly-Chaudière and Ghysen, 1986). In Drosophila, extensive research has shown that alternate segmental identities realized in the epidermis are controlled by the homeotic genes (Lewis, 1963, 1978; Mahaffey and Kaufman, 1987a and reviews by Akam et al., 1988 and Kaufman et al., 1989). Mutations in the homeotic genes have been shown to result in improper head involution and the transformation or “homeosis” of cuticular (epidermal) structures of one or more metameric segments towards the identity of another segment (Lewis, 1963, 1978; Sanchez-Herrero et al., 1985; Kaufman et al., 1989). The observation that there are distinct spatial patterns of sensilla in the embryonic PNS suggests that the homeotic genes may also be responsible for these differences.

There are two clusters of homeotic genes in the Drosophila genome: the Antennapedia complex (ANT-C) and the Bithorax complex (BX-C) (Kaufman et al., 1980, 1989; Lewis, 1978). The ANT-C contains the homeotic genes labial (lab), proboscipedia (pb), Deformed (Dfd), Sex combs reduced (Scr) and Antennapedia (Antp) which are required for head and anterior thoracic development, while the BX-C contains the homeotic genes Ultrabithorax (Ubx), abdominal-A (abd-A) and Abdominal-B (Abd-B) which specify posterior thoracic and abdominal identity. All of the homeotic genes are required for both adult and larval development with the exception of pb, which is required only for adult development (Pultz et al., 1988). The homeotic genes of Drosophila are believed to encode transcription factors containing a 60 amino acid motif called the homeodomain (McGinnis et al., 1984; Scott and Weiner, 1984) that has been shown to bind DNA in vitro (Müller et al., 1988; Desplan et al., 1985; Hoey and Levine, 1988). It has been proposed that the protein products of these genes elicit segmental identity through the activation of a genetic program involving downstream “realisator” genes which impart segmentspecific characteristics (García-Bellido, 1977).

If the specification of the PNS requires the action of the homeotics in a similar fashion to their role in the establishment of epidermal identity, an important question involves the time in development during which the homeotic genes influence PNS development. Moreover, the nature of the involvement of the homeotics in PNS specification is constrained by how the PNS develops relative to the known spatiotemporal patterns of homeotic gene expression. The formation of the Drosophila embryonic PNS from the ectoderm occurs between 5 and 9 hours of embryogenesis and is dependent on the action of several classes of genes (for a review, see Jan and Jan, 1990) such as the prepattem, proneural and neurogenic genes. The prepattern genes include the segmentation genes which provide positional information and divide the embryo into segmental units. The proneural genes include members of the achaete-scute complex (AS-C) and the daughterless (da) gene. These loci are expressed early in the ectoderm and control the expression of genes necessary for the commitment and differentiation of neuronal precursor cells (Ghysen and Dambly-Chaudière, 1989). Mutations in proneural genes result in a deficit of sensory organs (Dambly-Chaudière and Ghysen, 1987; Caudy et al., 1988). As a neuronal precursor cell develops, it inhibits its neighbors from becoming neuroblasts through the action of the neurogenic genes (Lehmann et al., 1983). This group includes Notch (TV)and Delta (DI), which encode transmembrane proteins with EGF-like repeats (Wharton et al., 1985; Kopczynski et al., 1988), and are probably involved in a cell-cell interaction pathway. Mutations in the neurogenic genes result in a hypertrophy of the PNS and CNS at the expense of the epidermis (Hartenstein and Campos-Ortega, 1986).

Mutations in the homeotic genes have also been demonstrated to result in defective PNS development. For example, previous reports describing defects in sensilla of the embryonic PNS for the homozygous mutant bithoraxoid (bxd) (Hartenstein, 1987) and effects of homeotic mutations on the central projections of sensory neurons in adult flies (Ghysen et al., 1983) provide evidence for a role in PNS development for the homeotic genes. Furthermore, the observation that homeotic gene products accumulate in ectodermal cells prior to the appearance of the first sensory progenitor cells (Ghysen and O’Kane, 1989) suggests that the homeotic genes may influence early events in PNS development. The goal of this study was to further define the roles of the homeotic genes in embryonic PNS development. This was accomplished by examining the embryonic PNS with cell specific markers in homozygous mutants for the homeotic genes Scr, Antp, Ubx, abd-A and Abd-B. In addition, ectopic expression of the homeotic genes lab, Dfd, Scr, Antp or Ubx via an hsp-70 promoter at various stages of development was employed to determine the effect of aberrant expression of these proteins on the normal development of sensory organs in the embryonic PNS.

Flies were raised at 25°C on standard medium supplemented with baker’s yeast. The wild-type Drosophila melanogaster stock used was Oregon R. The mutant stocks used were Scr⁴ red e/TM6B, which is a protein null for Scr (Wakimoto and Kaufman, 1981; Riley et al., 1987), Antp^w10 red e/TM6B, which accumulates no detectable Antp protein (Wakimoto and Kaufman, 1981; Carroll et al., 1986), Sb/In(3LR)Ubx¹⁰¹, a Ubx breakpoint mutation (Hayes et al., 1984), ahdA^M1/TM1, which is a protein null for abd-A (Sanchez-Herrero et al., 1985; Macias et al., 1990), Abd-B^MX2, which produces no detectable Abd-B protein in parasegments 10-12 (Sanchez-Herrero et al., 1985; Celniker et al., 1990), and Df(3R)Ubx¹⁰⁹/Dp(3,3)P5, Sb, which is deficient for the Ubx and abd-A transcription units (Lewis, 1978; Karch et al., 1985). Heterozygous parents were crossed to produce populations of embryos in which, on average, 25% of the animals were homozygous for the mutation of mterest. Between 25-50 homozygous mutant embryos were immunostained with each antibody for PNS analysis.

The transformant fine hsDfd58, which cames a Dfd cDNA under control of the hsp70 promoter, was kindly provided by W. McGinnis (Kuziora and McGinnis, 1988), while the transformant line hsUbxla22, which contains a Ubx cDNA under control of the hsp70 promoter, was kindly provided by D. Hogness (Mann and Hogness, 1990). The transformant hne hslab was generated by placing genomic fragments of the labial gene containing the coding sequences and second intron behind the hsp70 promoter in the P element vector CoSpeR Not I (S. Chouniard, personal communication) which contains the white gene as a visible marker (J Tamkin, personal communication). The transformant lines hsAntp and hsScr were generated by cloning the Antp cDNA G1100 (Scott et al., 1983) and an Scr cDNA (Mahaffey and Kaufman, 1987b) downstream of the hsp70 promoter in the P element vector CaSpeR (F. Randazzo, personal communication), which contains the white gene as a visible marker (Pirotta, 1988). Standard P-element mediated germline transformation techniques were used for obtaining independent insertion lines of the hslab, hsAntp and hsScr constructs (Robertson et al, 1988). The hslab line used in these experiments has the P-element insertion on the X chromosome and is homozygous viable The hsAntp line has the P-element insertion on the third chromosome and is balanced over TM3, while the hsScr line has the P-element insertion on the X chromosome and is homozygous viable.

The collection, fixation and staining of embryos was performed as described previously (Mahaffey and Kaufman, 1987b) Double staining of embryos was performed as described in Matthews et al (1990). The anti-Zab antibody was generated as described in Diedench et al (1989). The anti-Dfd and anti-Scr antibodies were generated as described in Mahaffey et al. (1989). The hybndoma hne 8C11.1 (anti-Antp) (J. Condie and D Brower, unpublished results) and the hybndoma hne FP3.38 (anti-i/bx) (White and Wilcox, 1984) were kindly provided by D. Brower and M. Muskavitch respectively. The supernatant from the hybridoma hne BP104 (anti-neuroglian) was kindly provided by C. Goodman and is described by Hortsch et al (1990). The F2 anti-Cut antibody was kindly provided by Y. Jan and is descnbed by Blochlinger et al. (1990). The anti-o’-tubuhn 85E antibody (P12a85-E) was kindly provided by K Matthews and is described by Matthews et al (1990). The antibodies 8C11.1, FP3.38 and BP104 were used as hybndoma supernatants at dilutions of 1:50,1:100 and 1:5 respectively. The other antibodies were affinity purified polyclonals derived from rabbits, with the exception of anti-Cut (rat). The anti-Zab, anti-Z)/d, anti-Scr, P12a-85E and F2 antibodies were used at dilutions of 1.150, 1:150,1:100, 1.100 and 1:500 respectively. The secondary antibodies used for immunostaining were goat anti-rabbit conjugated to HRP (Sigma) and goat anti-mouse (H+L) conjugated to HRP (Sigma). Both secondary antibodies were used at a dilution of 1:200 Stained embryos were examined and photographed using Nomarski optics with a Zeiss Axiophot Photographs were taken with Kodak Kodacolor Gold color print film with a blue filter in place for color correction and a tungsten light source.

Embryos were collected on hard agar plates coated with a dab of yeast paste for one hour at room temperature. A precollection of embryos was obtained to ensure that adult females did not retain fertilized eggs. Two different heat shock regimes were employed. In the first, embryos were allowed to age at 25°C for either 3, 4, 6 or 7 hours after the collection period, before administering a one hour heat shock at 37°C (35°C for hslab) by immersing the plates in a water bath of the appropriate temperature. After the heat shock, the embryos were allowed to recover at 25°C for one hour or until they were about 12-14 hours of development. In the second heat shock regime, embryos were allowed to age for 3 hours at 25°C, given a one hour heat shock at the appropriate temperature, allowed to recover at 25°C for one hour, given a second heat shock for 30 min, allowed to recover at 25°C for one hour, given a third heat shock for 30 min, and placed at 25°C until they were about 12-14 hours of development. For each of the regimes, approximately 25-50 embryos were examined with each antibody for PNS defects and 20-50 embryos were examined for cuticular transformations.

Following the heat shock period, embryos were allowed to age until about 24-26 hours of development and then washed several times The embryos were dechorionated in 50% bleach, followed by several washes, and then devitelhnized in heptane and methanol The embryos were then rehydrated in PBS, placed on a microscope slide and excess PBS was blotted away. A drop of polyvinyl lactophenol was applied to the embryos and a covershp was placed over them The shdes were then placed on a hot plate at 60°C for 24 hours. Cuticle preparations were examined and photographed under phase contrast optics on a Zeiss Axiophot.

The patterns of embryonic sensory organs were examined in homozygous mutant embryos for the homeotic genes Scr, Antp, Ubx, abd-A and Abd-B, and in embryos in which the homeotic genes lab, Dfd, Scr, Antp or Ubx were ectopically expressed via the hsp-70 promoter. In addition, denticle belts in the cuticle were examined in embryos in which the homeotic genes were ectopically expressed. Analysis of the PNS was accomplished with antibodies which recognize the proteins Cut (F2), neuroglian (BP104) and a-tubulin 85E (P12a-85E). These antibody markers allow for the detection of specific subsets of cells in the embryonic PNS. In order to identify definitively the PNS phenotype in protein null homeotic mutant embryos, double stains were performed with antibodies which recognize the homeotic proteins in conjunction with the antibodies which label cells in the PNS. For example, to identify the es organs in an Scr⁴ homozygous embryo, embryos derived from Scr⁴red e/TM6B parents were double immunostained with anti-Scr and anti-Cut antibodies. The anti-Cut staining pattern was then carefully examined in embryos which did not stain with anti-Scr antibodies. Since antibodies which recognize abd-A and Abd-B were not available, the mutant phenotype was taken as that present in one quarter of the embryos analyzed.

The F2 antibody recognizes the protein product of the cut gene, which is required for es organ development (Bodmer et al., 1987), and is localized to the nucleus of both neuronal and support cells of es organs, but not in cells of ch organs (Blochlinger et al., 1990). The monoclonal antibody BP104 recognizes a nervous system-specific form of neuroglian, a member of the immunoglobulin superfamily (Bieber et al., 1989), which is localized to the cell surface of neurons in the central nervous system (CNS) and neurons and some support cells in the PNS (Hortsch et al., 1990). The P12α-85E antibody specifically recognizes a cytoplasmic a-tubulin protein in the support cells of ch organs, but not es organs (Matthews et al., 1990). For a description of the embryonic expression patterns of cut, the nervous system-specific neuroglian and O--85E, see the papers by Blochlinger et al. (1990), Hortsch et al. (1990) and Matthews et al. (1990) respectively. The normal staining patterns for these antibodies are illustrated in Figs 1, 2A, 3A and 6A.

The normal pattern of sensory organs has been fully described elsewhere (Campos-Ortega and Hartenstein, 1985; Dambly-Chaudière and Ghysen, 1986; Hartenstein, 1988), thus only a brief description will be given here. There are three major types of sensory organs in the Drosophila embryo: es and ch organs described previously, and multiple dendrite (md) neurons, which are probably proprioceptive sensory cells that contain a single neuron with numerous dendrites and a single support cell. As mentioned previously, the sensory organs in the trunk are segmentally arranged according to five patterns (see Introduction). The segmental patterns of the sensory organs for the prothorax (Tl), meso- and metathorax (T) and abdominal (A) meta-meres are illustrated in Fig. 1. The majority of sensory organs, in any segment, reside in the anterior compartment, while the posterior compartment contains only ch organs laterally and dorsaUy (lch3 of Tl, dorsal triscolopidial organs (dch3) of T2 and T3, Ich5 of Al-A7) (Hartenstein, 1987). This study will focus on the ch and es organs in the Tl, T (T2,T3) and A (A1-A7) patterns only, as the sensory organs in the A8, A9 and head segments were difficult to identify. Within these segmental patterns, the cell bodies of the sensory organs are conspicuously arranged in four clusters from the most ventral to the most dorsal: v, v′, v″ (Tl only), 1 and d. Peripheral neurons are identified by: (1) their segmental pattern (T1-A9), (2) cluster (v-d), (3) nature (es, ch or md), and (4) position within a cluster from ventral to dorsal (A, B, C, D, E).

In a normal embryo, staining with the antibody P12a-85E reveals three sets of ch organs in the Tl pattern: a ventral monoscolopidial organ (vchl), a lch3 and a dch3 (Fig. 2A). Differences between the dch3 and lch3 in Tl include their location, axonal projections and number of attachment cells. While the axons from most ch organs in the embryo project into the segmental nerve of the segment in which they are located, the axons from the lch3 in Tl cross the segmental boundary and join with the T2 segmental nerve (Ghysen et al., 1986). There are two sets of ch organs in the T pattern: a vchl and a dch3 (Fig. 2A). The A pattern is characterized by ventral monoscolopidial organs A and B (vchA, vchB), a lateral monoscolopidial organ (Ichl) and a lch5 (Figs 2A, 4C). The es organs of the PNS are derived from neuronal precursors which are different from those giving rise to ch organs (Ghysen and O’Kane, 1989). Immunostaining of normal embryos with the F2 antibody reveals es organs present in three unique patterns in the trunk: Tl, T and A (Figs 1, 3A, B). The major differences between the Tl and T patterns are the presence of a v” cluster in Tl and the positions and types of es organs in both the lateral and dorsal clusters. The spatial distribution of es organs in the A1-A7 meta-meres is also easily distinguished from the patterns of es organs in the thoracic segments (Fig. 1, 3A, B).

Loss of function mutations in the homeotic genes result in the transformation of some sensilla and the development of ectopic sensilla as revealed with the P12a85E, BP104 and F2 antibodies. Similar to certain cuticular defects observed in homeotic mutant embryos, defects in sensilla also show variable penetrance and expressivity, thus only general descriptions are given here. The effects of homeotic mutations on the normal patterns of sensilla are summarized in Table 1.

Previous studies have indicated that in the absence of Scr function, Tl is transformed to a more posterior thoracic identity (Wakimoto and Kaufman, 1981; Sato et al., 1985; Pattatucci et al., 1991). The defects in sensilla of Scr⁴ embryos .follow this type of transformation (Figs 2B, 3C,D). Interestingly, while there is 100% transformation of the lch3 in Tl to a dch3, in about 50% of Scr⁴ embryos the axonal projection of this transformed sensory organ still crosses the T1/T2 segmental boundary and joins the axons of the T2 segmental nerve as in the wild-type embryo (data not shown).

In Antp^w10 embryos, cuticular features indicate that T2 and T3 are transformed towards a more anterior thoracic identity (Wakimoto and Kaufman, 1981). However, the defects in sensilla of Antp^w10 embryos do not follow this type of transformation. Instead, there is no apparent homeotic transformation, yet there is abnormal development of many sensilla in all the thoracic segments. Specifically, the positions of the dch3 organs in T2 and T3 with respect to the cuticle are slightly abnormal in about 50% of mutants and ectopic monoscolopidia frequently appear in ventral positions in T2, T3 or both (Fig. 2C). A particularly interesting observation is that while the position of the lch3 in Tl is normal in Antp^w10 embryos, double immunostaining with P12o85E and BP104 reveals that the axons from the lch3 do not cross over to T2, but appear to join the Tl segmental nerve (data not shown). The es organs of the thoracic segments are also affected by the loss of Antp function. The anterior spiracle is dramatically reduced and malformed, and Keilins organs are incomplete in all three segments (Fig. 3E, F). There are losses of es organs in the v′, 1, and d clusters of all three thoracic segments, but the T2 segment is most severely affected in all cases (Fig. 3E, F).

The loss of Ubx function in the embryo has been previously shown to result in a transformation of anterior T3 and Al towards a T2 identity and posterior T2 and T3 towards a Tl identity (Hayes et al., 1984; Sanchez-Herrero et al., 1985). The defects in sensilla observed in Ubx¹⁰¹ embryos also follow this type of transformation (Figs 2D, 4A). Curiously, as observed in Antp^w10 embryos, many Ubx¹⁰¹ embryos also exhibit ectopic monoscolopidial organs located in a ventral position m T3, Al or both.

In abd-A^M1 embryos, previous studies have indicated that the A2-A7 metameres are transformed toward an Al identity (Sanchez-Herrero et al., 1985). PNS defects in abd-A loss of function embryos follow this type of transformation for most sensilla in the abdominal segments (Figs 2E, 4D). However, the lch5 organs of the abdominal segments are transformed more anteriorly to a posterior thoracic identity (Figs 2E, 4D).

The loss of Abd-B function in Abd-B^MX2 embryos has been shown to result in a transformation of A5-A7 towards an A4 identity (Sanchez-Herrero et al., 1985). There are no detectable defects in the sensilla of these embryos (data not shown). Nevertheless, it is possible that a similar transformation occurs in the PNS, since in normal embryos the sensilla in A5-A7 are indistinguishable from those in A4. Previous studies have also shown that the identity of the abdominal segments is transformed to a thoracic identity in Ubx¹⁰⁹ embryos (Lewis, 1978) which are deficient in the Ubx and abd-A transcription units (Karch et al., 1985). Defects in the sensilla of these embryos follow the same type of transformation (Figs 2F, 4B).

The homeotic genes lab, Dfd, Scr, Antp or Ubx were ubiquitously expressed in embryos under the control of the hsp 70 promoter by administering a heat shock at various times in development (see Materials and methods). As a way of verifying that the fly fines carrying these constructs were not subject to leaky expression of these genes, embryos were collected over a 16 hour time period without the administration of a heat shock and immunostained with the appropriate anti-homeotic protein antibody (i.e., hslab with the anti-Zah antibody, hsDfd with the anti-D/d antibody, etc…). All the hs lines exhibited normal patterns of immunostaining with the exception of the hslab and hsUbx lines, in which a small percentage of embryos showed some ectopic immunostaining (data not shown). However, the percentage of hslab and hsUbx embryos exhibiting aberrant expression is too small to account for the phenotypes observed following heat induction. As a further control, cuticles and the PNS were examined in non-heat shocked embryos from all the hs lines and were found to be normal (data not shown). To determine whether ubiquitous expression of the homeotic protein was induced following a heat shock, embryos were administered a one hour heat shock, and then immunostained with the appropriate anti-homeotic protein antibody 30 minutes and one hour after the heat shock. In all the hs lines, the appropriate homeotic protein was detected in all cells of the embryo at both 30 minutes and one hour post heat shock (data not shown).

In order to confirm that the hs lines employed in this study exhibit the same cuticular phenotypes following heat induction that have been previously described (hsDfd, Kuziora and McGinnis, 1988; hsAntp, Gibson and Gehring, 1988; hsScr, Gibson et al., 1990; hsUbx, González-Reyes and Morata, 1990; Mann and Hogness, 1990), embryos were administered a one hour heat shock at various stages of development and cuticles were prepared for microscopy (described in Materials and methods). The cuticular transformations resulting from overexpression of the homeotic proteins are summarized in Table 2. The cuticular phenotypes observed for hsDfd, hsScr, hsAntp, and hsUbx are in agreement with the results previously reported (Fig. 5C-F). Since the phenotype of hslab has not been reported as yet, a description will be presented here.

During embryogenesis, the lab protein is normally expressed in epidermal cells of the procephalon in the intercalary segment, dorsal ridge of the head and also in endoderm cells of the midgut (Diederich et al., 1989). The hslab construct is unique in that it contains an intron, thus heat shocks were administered at 35°C to ensure that proper splicing of the RNA product was efficient (Yost et al., 1990). The cuticular phenotype of an embryo in which the lab protein was ectopically expressed is illustrated in Fig. 5B. The effects of aberrant lab expression include disruption of normal head involution and defects in the cephalopharyngeal skeleton. In particular, the lateralgrâte is dramatically reduced or missing entirely and the dorsal bridge, dorsal arms, ventral plate, ventral arms and T-bars are malformed. The epistomal sclerite, ectostomal sclerite, labral sense organ and labrum are sometimes missing and sclerotized material is occasionally observed around the dorsal pouch area. While these structures are either missing or malformed, there are no apparent homeotic transformations. Four different heat shock regimes were employed (see Materials and methods) to determine the stage of development in which cuticular structures are most sensitive to aberrant expression of the lab protein (Table 2). The multiple heat shock regime seemed to give the most severe defects, while defects in the cephalopharyngeal skeleton were most common when the heat shock was administered at 4 hours after egg lay (AEL).

As ectopic expression of the homeotic genes lab, Dfd, Scr, Antp and Ubx has been demonstrated to cause abnormalities or transformations in the segmental identities of the embryonic epidermis, it was of interest to determine if ubiquitous expression of these proteins would lead to the same types of transformations in the PNS. To answer this question, embryos were collected from the hs lines, administered a one hour heat shock at various times in development, allowed to recover until the embryos were about 12-14 hours old, and then the PNS was analyzed with the F2, BP104 and P12a-85E antibodies. The data from these experiments are summarized in Table 2. As there were numerous abnormalities in the PNS of all the hs lines, and the phenotypes varied from one embryo to another, only a general description of the most consistent defects will be presented here. Major defects were observed in both es and ch organs of the PNS in all the hs lines when the heat shock was administered at 3 or 4 hours AEL, while heat shocks administered at 6 or 7 hours AEL caused only minor (mostly in the thoracic ch organs) or no abnormalities (Table 2). PNS abnormalities were never observed in any of the heat shocked Oregon R embryos, which were used as a control in all experiments.

Ectopic expression of Ubx has been reported to transform the identity of the thoracic and head segments towards that of an Al identity (González-Reyes and Morata, 1990; Mann and Hogness, 1990). One of these studies also demonstrated that the PNS of a hsUbx animal also follows this type of transformation (Mann and Hogness, 1990). We used the same hsUbx line in this study with similar results (Fig. 6B, C). However, in addition, we find some defects in sensilla of the thoracic segments which indicate a partial transformation of these segments toward an anterior thoracic identity. For instance, the dch3 organs in T2 and T3 are not transformed to a lch5 typical of the abdominal segments, but are partially transformed to a prothoracic lch3 (Fig. 6B). Interestingly, while the position and morphology of this sensory organ are characteristic of a prothoracic lch3, the axonal projection is not (i.e., it does not project to the next segmental nerve).

As ectopic expression of Ubx protein has been shown to alter segmental identity only in segments anterior to its normal domain of expression, previous studies have also indicated that this rule applies for ectopic expression of the Antp protein (Schneuwly et al., 1987; Gibson and Gehring, 1988). These studies demonstrated that ubiquitous expression of the Antp protein causes a failure in head involution and a clear transformation of prothoracic identity towards mesothoracic identity. The PNS defects observed in hsAntp embryos do not show any parallel homeotic transformations. However, while most sensilla in the abdominal segments are not affected, the lch5 organs in Al and A2 are frequently transformed to triscolopidial or tetrascolopidial organs (Fig. 7H). In addition, there is no clear transformation of sensilla in Tl towards a T2 pattern, but rather the sensilla appear in abnormal numbers and positions with respect to the cuticle (Figs 7G, H, 8D, F).

While hsAntp embryos show a cuticular transformation of prothoracic identity towards mesothoracic identity (Gibson and Gehring, 1988), the reciprocal transformation occurs in hsScr embryos, namely, posterior thoracic identity is transformed to a prothoracic identity (Gibson et al., 1990). The PNS phenotype in hsScr embryos doesn’t follow this type of transformation and reveals no obvious parallel homeotic transformation. Despite this fact, there are discernable defects in sensilla in all segments of the trunk (Figs 7E, F, 8B), indicating that ubiquitious expression of the Scr protein can alter the fates of cells in the abdominal segments, an observation at odds with previous conclusions (Gibson et al., 1990). The most dramatic transformation of sensilla occurs in the abdominal segments where the lch5 organs are transformed into mono- or discolopidial organs (Figs 7F, 8B).

Similar to hsScr embryos, which show a posterior directed transformation in the cuticle, hsDfd embryos also exhibit a posterior directed transformation. That is, a transformation of the thoracic segments toward maxillary identity is observed (Kuziora and McGinnis, 1988). There is no apparent homeotic transformation in the PNS, as again sensory organ defects do not follow the transformation seen in the cuticle. However, PNS defects are observed in hsDfd embryos and these are very similar to those defects seen in hsScr embryos (Figs 7C, D, 8E). Sensilla in hsDfd embryos occur in abnormal numbers and positions with respect to the cuticle in all segments of the trunk. Curiously, the dch3 organs in T2 and T3 are very sensitive to overexpression of the Ubx, Scr and Dfd products, exhibiting a transformation to a prothoracic lch3. However, this transformation is incomplete, as the axonal projection of these lch3 organs is not the same as in wild-type embryos (data not shown).

While previous studies have shown that overexpression of the Ubx, Antp, Scr and Dfd proteins results in cuticular transformations in the embryo, this study has demonstrated that this is not the case for the lab protein (Fig. 5B). As there is no apparent homeotic transformation in the cuticle of hslab embryos, the same is true for the PNS in these embryos. Overexpression of the lab protein does, however, cause PNS defects in all segments of the trunk, with a general reduction in the number of sensory organs and alterations in their normal spatial patterns (Figs 7A, B, 8C). In addition, hslab embryos frequently display a partial transformation of the dch3 organs in T2 and T3 to a prothoracic lch3 similar to that seen in embryos overexpressing the Ubx, Scr and Dfd products.

While there are no obvious homeotic transformations in the PNS which parallel observed cuticular changes accompanying overexpression of the lab, Dfd, Scr and Antp proteins, sensory organs are often found in abnormal numbers and locations Could ectopic expression of homeotic proteins, in general, act as a switch in neuronal precursor development and transform es organ precursors into ch organ precursors or vice versa? In order to answer this question, heat shocked embryos were double stained with the P12a-85E and F2 antibodies to detect both es and ch organs in the same embryo (Fig. 8). Examination of these embryos with the P12a--85E and F2 antibodies indicates that transformations between es and ch organ precursors is not a likely cause of the abnormalities observed in the sensory organs.

The homeotic genes Scr, Antp, Ubx, abd-A and Abd-B encode protein products which provide positional information in an antero-postenor fashion to cells in the epidermis, by imposing segment-specific cuticular characteristics (Akam et al., 1988; Kaufman et al., 1989). In the absence of the various homeotic gene products, the epidermal characteristics of one or more segments are transformed into the characteristics of another segment. The homeotic proteins are expressed in a complex manner during embryogenesis, in a majority of cases beginning around the time of germ band extension. In the ectoderm, the homeotic genes are, in general, expressed in parasegments (PS) which consist of the posterior compartment of one segment and the anterior compartment of the next most posterior segment. The most anteriorly accumulated homeotic proteins are lab, Dfd and pb, which are present in cells of the head segment primordia (Diederich et al., 1989; Regulski et al., 1987;Pultzet al., 1988). The Scr protein accumulates in cells of the labial and Tl segments (Riley et al., 1987; Mahaffey et al., 1989), while the Antp protein is localized in cells from the posterior compartment of the labial segment to the anterior compartment of Al (PS 3-6), with highest levels in PS 4 (Carroll et al., 1986). Ubx protein appears in a complicated pattern from PS 5-12, with highest levels in PS 6 (White and Wilcox, 1984; Carroll et al., 1988). The expression of the abd-A protein overlaps with the Ubx pattern and is detectable in cells from PS 7-13, with the highest levels in the posterior compartments of these parasegments (Karch et al., 1990; Macias et al., 1990). The homeotic protein expressed most posteriorly is the Abd-B protein, which accumulates in cells of PS10-15, with highest levels of protein in PS 13-15 (Celmker et al., 1989; DeLorenzi and Bienz, 1990).

There are also studies which indicate that some of the homeotic proteins are expressed in a subset of cells of the PNS at later stages of development. For instance, the Antp (Carroll et al., 1986), lab (Diederich et al., 1989), Ubx and abd-A proteins (Karch et al., 1990) have been localized to cells of the PNS in the late embryo. However, the function of this later expression is unknown as cells in the PNS have already formed and differentiated at this time. Possibly, the homeotic proteins are involved in maintaining the differentiated state of these cells by regulating such processes as neurotransmitter synthesis. The present study was aimed at determining the effects, if any, of loss of function and gain of function mutations of the homeotic genes on the generation of the segment-specific patterns of sensilla in the PNS. Our results suggest that the homeotic genes, through their early expression in the ectoderm, do provide positional information which is necessary for the appropriate development of the segmental spatial patterns of sensilla in the trunk of the embryo.

The PNS was examined in homozygous mutant embryos carrying null alleles for the homeotic genes Scr, Antp, Ubx, abd-A and abd-B. We utilized antibody markers which allowed for the identification of both ch and es organs in the PNS. Our results indicate that, m general, loss of homeotic gene function results in a similar type of transformation in the PNS as observed in the epidermis. Perhaps, this result is not too surprising when one considers that both epidermal structures and sensory organs are derived from the same ectodermal tissue and may respond to the same or similar positional cues. However, there are exceptions to the rule, which suggests that the homeotic genes may act in concert to generate some sensilla, and act autonomously to generate others. On the other hand, the results point out that not all sensilla are affected by loss of function mutations in the homeotic genes, supporting the view that many sensilla are probably determined by the action of the segmentation genes (Hartenstein, 1987).

The lch3 organ in the prothoracic segment is one sensory organ which appears to be affected by the action of two different homeotic gene products, Scr and Antp. This sensory organ is a derivative of the posterior compartment of Tl (Hartenstein, 1987) and is different from other ch organs in the embryo in that it projects its axons from Tl to the T2 segmental nerve (Ghysen et al., 1986). Loss of function of Scr results in a transformation of this organ to a dch3, an organ found in the T2 and T3 segments. However, in about 50% of these embryos, the axons still cross the segmental boundary and join the T2 segmental nerve. Furthermore, in Antp loss of function embryos, the lch3 in Tl is morphologically normal, yet the axonal projection is abnormal. Thus it would appear that Scr dictates the morphology and position of the prothoracic lch3, while Antp determines the axonal projection. Interestingly, Scr protein accumulates in cells of the posterior compartment of Tl at lower levels than in cells of the anterior compartment of Tl (Mahaffey and Kaufman, 1989). However, as the absence of Scr function leads to a change in the identity of the lch3, which is a derivative of the posterior compartment (Hartenstein, 1987), we can conclude that the low level of Scr expression in cells of the posterior compartment of Tl is significant, at least for the specification of this one structure.

Defects in sensilla observed in Antp loss of function embryos also do not follow a simple homeotic transformation. While previous studies have demonstrated a transformation of T2 and T3 towards Tl identity in Antp^w10 embryos (Wakimoto and Kaufman, 1981), this type of transformation is not observed in the PNS. Instead, there is a dramatic reduction in the number of es organs in the thoracic segments. Compared to the es organs, the ch organs of the thoracic segments are relatively unaffected in most Antp^w10 embryos, yet there are ectopic monoscolopidial organs which develop ventrally in T2 and T3. This phenotype is reminiscent of the patterns of sensilla characteristic of the head segments, which contain relatively few es organs and only a single monosclopidial organ per segment (Hartenstein and Campos-Ortega, 1986). This would seem to indicate a partial transformation of the PNS in the thoracic segments towards head identity in Antp^w10 embryos. This interpretation is strengthened by the observation that structures resembling mouthparts develop in the thoracic segments of embryos carrying strong loss of function alleles of the Antp gene (T. Kaufman, unpublished observations).

The PNS defects observed in loss of function mutants for the homeotic genes which specify the identity of the abdominal segments, again suggest that the homeotic genes function both independently and together to influence sensilla development in the abdomen. The Abd-B^MX2 mutant is a protein null for PS10-12 (A5-A7) (Celniker et al., 1990) and showed no obvious defects in the sensilla of these segments. This result indicates that the expression of Abd-B in A5-A7 is apparently unecessary for sensilla formation, although it has a known role in establishing the cuticular identity of these segments. While Abd-B has no apparent role in sensilla development of the A pattern, either Ubx or abd-A function alone appears to be sufficient for the proper development of most sensilla in A1-A7, strengthening the argument that some functions of Ubx and abd-A are redundant where their patterns of expression overlap (Akam et al., 1988). However, Ubx and abd-A are also required independently for the formation of specific sensilla in Al-A7. For instance, the formation of the lch5 organs of A1-A7 are dependent .on abd-A function. The transformation of this organ to a dch3 in abd-A^MI embryos has been previously reported (Karch et al., 1990) and is probably due to increased expression of Ubx protein in the posterior compartments of the abdominal segments of abd-A^MI embryos (Macias et al., 1990). Ubx, on the other hand is necessary for the formation of all sensilla in Al which are derivatives of the anterior compartment, such as all the es organs, the Ichl, vchA and vchB (Hartenstein, 1987).

Previous results have shown that ectopic expression of Dfd (Kuziora and McGinnis, 1988), Scr (Gibson et al., 1990), Antp (Gibson and Gehring, 1988), and Ubx (González-Reyes and Morata, 1990; Mann and Hogness, 1990) via an hsp-70 promoter causes homeotic transformations of cuticular structures in the head and thorax. However, ectopic expression of the homeotic gene lab doesn’t result in any apparent homeotic transformation, but rather causes defective head involution and abnormal development of the cephalopharyngeal skeleton. These latter types of defects are also commonly observed in embryos in which the Dfd, Scr, Antp and Ubx genes are overexpressed, although with much more severity. The structures most affected by overexpression of lab are the lateralgrate, ventral arms of the mandibular segment and the dorsal bridge, dorsal arms and vertical plates of the acronal segment. Ectopic expression of lab can therefore influence the normal development of structures derived from segments both anterior and posterior to the normal domain of lab expression, similar to the effects observed in embryos overexpressing the Dfd and Scr genes (Kuziora and McGinnis, 1988; Gibson et al., 1990). However, the absence of any apparent homeotic transformation in hslab embryos stands in contrast to results obtained from the overexpression of other homeotic genes which cause specific homeotic transformations. This would seem to indicate that the presence of the lab protein is either masked in cells which express other homeotic proteins or that the lab protein is extremely unstable in cells outside its normal domain of expression and thus cannot exert any effect. In keeping with this latter possibility, the lab protein does appear to be more unstable than the other homeotic proteins when ectopically expressed, as the protein is barely detectable at one hour after heat shock (J. Heuer, unpubfished observations). Alternatively, the lab protein may not effectively compete with other homeotic proteins for downstream genes or maybe the genetic program normally under control of the lab protein in the epidermis is closer to the so called “ground state” than the other more posterior acting homeotic genes. If the latter were true, then overexpression of lab would not dramatically interfere with the development of segmental characteristics imposed by the other homeotic genes.

While the cuticular effects of overexpressing the lab protein are limited to the acronal, mandibular and labial segments, defects in the PNS are observed in all the thoracic segments and in some of the abdominal segments. Major defects in sensilla are also observed m the abdominal segments in which the Dfd and Scr proteins are overexpressed. This contrasts with previous studies which have concluded that the identity of the abdominal segments is unaffected by overexpression of the Dfd and Scr proteins (Kuziora and McGinnis, 1988; Gibson et al., 1990). In addition, while ubiquitous expression of Dfd, Scr or Antp protein causes obvious homeotic transformations in epidermal structures, similar conclusions cannot be drawn for the PNS. The phenotypes of the PNS in embryos overexpressing the lab, Dfd and Scr gene products are very similar. These embryos have reduced numbers of sensilla in both the thoracic and abdominal segments relative to the wild-type embryo. In addition, the lch5 organs of the abdominal segments are transformed to mono-or discolopidial organs. As both of these phenotypes are characteristic of sensilla in the head segments, we cannot exclude the possibility that overexpression of the lab, Dfd or Scr products results in a transformation of the PNS toward head identity, similar to that observed in Antp^w10 embryos. However, an alternate interpretation is that the presence of such high levels of head specific homeotic proteins expressed in the trunk segments nonspecifically disturbs the positional information normally required by PNS precursor cells to form the proper sensilla. The observed morphological abnormalities would result from a reduction in the number of cells recruited to form sensory organs in the thorax and abdomen rather than a homeotic transformation.

Could the PNS defects which occur in these embryos simply be caused by overexpression of any homeodomain containing protein, in that high levels of these proteins in the nucleus might lead to aberrant gene regulation? This possibility seems unlikely, as the cut gene product, which also contains a homeodomain, causes a much different, yet unique phenotype in the PNS following ubiquitous expression upon heat induction (Blochlinger et al., 1991). In addition, while overexpression of the lab, Dfd and Scr products causes major PNS defects in the abdominal segments, overexpression of the Antp and Ubx products has relatively little or no effect on the PNS in these segments.

The observation that both loss of function and gain of function mutations in the homeotic genes cause aberrant PNS development leads us to ask about the mechanism by which these gene products might influence this process. Clearly, the results of the heat shock experiments demonstrate that alterations in the sensilla patterns are more susceptible when the homeotic proteins are overexpressed at 3-4 hours of development compared to 6-7 hours of development. This result suggests that the homeotic genes likely function to influence sensilla patterns at an early stage of PNS development, possibly during or before the first sensory precursors arise, which is around 5 hours of development (Ghysen and O’Kane, 1989). This time period coincides with a time in PNS development when the proneural and neurogenic genes are presumbably active. As the homeotic gene products are nuclear proteins with the potential to act as transcriptional regulators, the possiblility exists that the homeotic proteins influence the regulation/expression of these genes. Our results in this study provide several lines of evidence which support this conclusion, albeit, indirectly: 1) sensory organs in homeotic loss of function or gain of function mutant embryos always consist of the proper number of neuronal and support cells, indicating that cell lineage genes are not affected; 2) sensory organs are not transformed from one type of organ into another, thus neuronal type selector genes are not affected; 3) the PNS phenotypes of loss of function or gain of function mutant embryos are similar to the PNS phenotypes observed in loss of function mutants for the proneural or neurogenic genes (i.e., too few or too many of one type of sensory organ). While we propose that the homeotic gene products affect sensilla patterns through the regulation of early events in PNS development, we cannot rule out the possibility that the homeotic genes might also affect the processes of cell death or cell division in the PNS, as we examined the PNS àt only one stage in development (12-14 hours). Nevertheless, the possibility that the homeotic proteins might regulate the genes which are necessary for the generation of sensory progenitor cells, either directly or indirectly, certainly warrants further studies aimed at addressing this question.

We would like to thank Yuh Nung Jan for sending us the F2 antibody, Mark Seeger and Corey Goodman for sending the BP104 antibody, Kathy Matthews for the P12a--85E antibody, Corey Goodman and Marc Muskavitch for the Ubx antibody, Danny Brower for the Antp antibody, and Marie Mazzulla for the preparation of the lab, Dfd and Scr antibodies. We would also like to thank the Bloomington stock center for supplying us with mutant fly lines, David Hogness for the hsUbx line, William McGinnis for the hsDfd hne, Scott Chouinard for the hslab hne and Fihppo Randazzo for the hsScr and hsAntp lines. We are grateful to Mark Seeger for sharing his thoughts with us on this project, to Scott Chouinard and Angela Pattatucci for advice on immunostaining, and to Kathy Matthews, Filippo Randazzo, Maureen Gorman and Scott Chouinard for critical reading of the manuscript We would also like to thank an anonymous reviewer, whose comments were very helpful in the revision of this manuscript. Finally, we are indebted to Dee Verostko for her expert administrative assistance. This work was supported by the Howard Hughes Medical Institute. J.G.H. is an associate for the Howard Hughes Medical Institute. T.C.K. is an investigator of the Howard Hughes Medical Institute. This work was also supported by NIH grant GM4299 to T.C.K