Charlatan, a Zn-finger transcription factor, establishes a novel level of regulation of the proneural achaete/scute genes of Drosophila

A classical example of two-dimensional pattern is that formed by the bristles and other types of sensory organs (SOs) in the epidermis of the adult Drosophila fly (Lindsley and Zimm, 1992). On the head and the dorsal mesothorax (notum),conspicuous large bristles (macrochaetae) arise in stereotyped positions,while smaller bristles (microchaetae) appear in density patterns. During the third instar larval and early pupal stages, the location of each macrochaeta is specified by the emergence of a precursor cell (SO precursor cell, SOP) at a stereotyped position of the imaginal disks, the larval epithelia that give rise to a large part of the adult epidermis(Cubas et al., 1991; Huang et al., 1991). This accurate positioning of SOPs in the imaginal disks is thought to be the culmination of a multistep process in which positional information is gradually refined (reviewed by Gómez-Skarmeta et al.,2003; Modolell and Campuzano,1998).

A key step of this process is the expression of the proneural genes achaete (ac) and scute (sc) in groups of cells, the proneural clusters, that prefigure the sites of the future macrochaetae (Cubas et al.,1991; Romani et al.,1989; Skeath and Carroll,1991). These genes, members of the achaete-scute complex(ASC) (reviewed by Campuzano and Modolell,1992; Ghysen and Dambly-Chaudière, 1988; Ghysen and Dambly-Chaudière,1989), encode transcriptional factors of the basic helix-loop-helix (bHLH) family. These factors confer to cells the potential to become SOPs, presumably by implementing neural differentiation programs. From each proneural cluster, a fixed number of SOPs are born, usually one or two. The proneural clusters of the wing imaginal disks (the precursors of each heminotum, wing and mesothoracic pleura) not only appear in constant positions, but each of them has a characteristic size, shape and time of appearance and disappearance (Cubas et al., 1991; Skeath and Carroll,1991). Moreover, a typical cluster that gives rise to one bristle may consist of 20 to 30 cells, but the SOP is selected from a smaller subgroup of cells that accumulate higher levels of Ac-Sc proteins than their neighbors,which constitute the proneural field(Cubas et al., 1991; Cubas and Modolell, 1992; Skeath and Carroll, 1991). This subgroup and the SOP, which accumulates the highest levels of Ac-Sc,always occupy the same position within the cluster. Hence, the expression of ac/sc in proneural clusters is exquisitely regulated.

The regulation of ac/sc is effected by means of two classes of cis-regulatory sequences, namely, cluster-specific and SOP-specific enhancers. The first type normally directs expression of both ac and sc in one specific proneural cluster and defines many of its characteristics, such as position, size and shape. These cluster-specific enhancers appear to be controlled by local combinations of transcription factors that together form a prepattern (reviewed by Ghysen and Dambly-Chaudière,1988; Gómez-Skarmeta et al., 2003). Expression occurs only at sites with the appropriate combinations of factors. Although in a few cases some of the prepattern factors have been identified, most of them remain unknown. Moreover, we still lack a clear understanding of how the inputs of the prepatterning factors are integrated into the patterns of proneural gene expression characteristic of each cluster.

The second type of enhancer mediates the strong expression of proneural genes in SOPs (Culí and Modolell,1998) by allowing self-stimulatory loops of expression of ac,sc and asense (ase), another bHLH member of the ASC(Brand et al., 1993; Domínguez and Campuzano,1993; Jarman et al.,1993a). The activation of these loops in one of the cells of the proneural field is an early and essential step of SOP commitment. This loop is also dependent on the presence of the Senseless (Sens) protein(Jafar-Nejad et al., 2003). The SOP-specific enhancers are also the targets of the inhibitory interactions that occur within the cells of the proneural cluster mediated by the Notch signaling pathway via E(spl) proteins(Culí and Modolell,1998; Giagtzoglou et al.,2003). By antagonizing these enhancers, N signaling, activated by Ac/Sc in the cells of the cluster, maintains them in a non-SOP state (mutual inhibition) (reviewed by Artavanis-Tsakonas et al., 1995). However, in a little-understood process, one cell of the proneural field escapes this inhibition, starts the proneural self-stimulatory loop and becomes an SOP. The developing SOP then signals via Notch in order to impede the remaining cells of the field from becoming SOPs(lateral inhibition) (Heitzler and Simpson, 1991; Simpson,1990; Simpson,1997). These SOP-specific enhancers are also the targets of positive interactions between the cells of proneural clusters mediated by the EGFR, which is necessary for the emergence of the SOPs of the notum macrochaetae (lateral cooperation)(Culí et al., 2001). To prevent the determination of excess SOPs from a proneural cluster, the levels of EGFR signaling must be regulated. This event seems to be accomplished in part by a negative effect on EGFR signaling of the N-mediated interactions that occur among cells of the proneural cluster.

The ac, sc and ase genes are also necessary for the formation of the external SOs of embryos and larvae(Dambly-Chaudière and Ghysen,1987). The process is similar to that in the imaginal disks(Ruiz-Gómez and Ghysen,1993). Other proneural genes are responsible for the development of the internal chordotonal organs (atonal)(Jarman et al., 1993b) and other neurons of the larval peripheral nervous system (PNS) (amos)(Huang et al., 2000; Villa-Cuesta et al.,2003).

Here we report the identification of a novel gene, charlatan(chn), which is involved in the development of the adult pattern of macrochaetae. chn defines a new level of control of ac/scthat is intermediate between the prepattern genes and the ac/scself-stimulation mediated by the SOP-specific enhancers. Thus, chn,which encodes a zinc finger transcription factor, is activated by ac/sc in the proneural clusters of the wing disk. In turn, chn stimulates the expression of ac/sc in these clusters. This enhanced expression facilitates the formation of SOPs. Our data indicate that the Chn protein reinforces the expression of ac/sc by acting,probably directly, on the proneural cluster-specific enhancers of the ASC. chn is also required for correct development of the embryonic/larval PNS, as its absence removes neurons and causes malformations of chordotonal organs.

The Drosophila stocks used were: ln(1)sc^10.1and Dp(1,2)sc¹⁹(Lindsley and Zimm, 1992), Ax^M1(Díaz-Benjumea and García-Bellido, 1990), pnr^VX6 and pnr^V1 (Heitzler et al., 1996), y w FLP122; act-FRT y⁺FRT-Gal4 UAS-GFP/SM6a-TM6b Tb(Ito et al., 1997), y w FLP122; P[ubiGFP] FRT42D/CyO, y w FLP122 f^36A; ck Pf[+] FRT42D/CyO, y w FLP122;y⁺ FRT42D/CyO (FlyBase). C765-Gal4(Gómez-Skarmeta et al.,1996), MS1096-Gal4(Lunde et al., 1998), eyg-Gal4 (Aldaz et al.,2003), MS248-Gal4(Sánchez et al., 1997), en-Gal4 y ap-Gal4 (Calleja et al.,1996), pnr-Gal4(Heitzler et al., 1996), dpp-Gal4 (Staehling-Hampton et al., 1994), sca-Gal4(Hinz et al., 1994), 69-B-Gal4 (Brand and Perrimon,1993), NP1212 (GETDB, Gal4 Enhancer Trap Insertion Database), UAS-sc and UAS-ato(Parras et al., 1996), AS1.4DC-lacZ(García-García et al.,1999), ANP-A-lacZ (M. J. García-García, PhD thesis, Universidad Autónoma de Madrid, 1999), 3.7-lacZ(Gómez-Skarmeta et al.,1995), 2.3-lacZ(Culí, 1998), SRV-lacZ (Culí and Modolell, 1998), chn^42/18(Kania et al., 1995), and chn^ECJ1, UAS-chn, UAS-chni, UAS-bda, this study.

UAS-chn was prepared by subcloning the cDNA of CG11798from clone SD05496 (BDGP) into the pUASt vector(Brand and Perrimon, 1993). UAS-bda was prepared by subcloning a PCR product containing the entire ORF of bda that was obtained using genomic Oregon R DNA as template. The PCR product was sequenced to confirm the fidelity of amplification. To prepare UAS-chni, a 400 bp fragment of chncDNA was amplified by PCR using 5′-GGGATCCCAAGCGGCTGCAGCTGC-3′upper primer and 5′-TGGAAGCTTCAACTCGTGCACGCC-3′ lower primer. The PCR product was cloned as a BamHI-KpnI fragment in the pHIBS vector (Nagel et al., 2002),to make the pHIBS-chn construct. A BamHI-SacI fragment from pHIBS-chn was subcloned in the pBluescript vector to generate pBS-chn. A KpnI-SalI fragment from pHIBS-chn and a KpnI-SalI fragment from pBS-chn were cloned in opposite directions in the pUASt vector, thus forming the final RNAi construct. All the UAS constructs were injected into y w embryos to obtain transgenic Drosophila lines by standard procedures(Rubin and Spradling,1982).

All overexpression experiments were carried out at 25°C, except for the flies shown in Fig. 1D,E,F,which were raised at 18°C, and those shown in Fig. 1G,H,I, which were cultured at 18°C, shifted to 25°C at 72 hours after egg laying, and returned to 18°C at puparium formation.

Males carrying the P element l(2)42/18(chn^42/18/CyO) were crossed to females carrying theΔ2-3 transposase (Cooley et al.,1988). Excisions of the l(2)42/18 transposon were selected by the loss of the w⁺ eye marker in the F1 progeny. Individuals were crossed to w; If/CyO flies and balanced. One hundred lines that failed to complement the original mutation were selected. Lines with the strongest defects in the embryo PNS, as detected by staining with mAb 22c10, were selected and the presence of chn mRNA was examined. chn^ECJ1 mutant embryos lacked this mRNA. To identify its molecular lesion, genomic DNA, prepared from homozygous chn^ECJ1 mutant larvae, was used as a template in PCR reactions to amplify the genomic region near the insertion point of the P element l(2)42/18.

To generate clones of cells mutant for chn, either y w hs-FLP122, f^36a; ck Pf[+] FRT42D/CyO or y w hs-FLP122; P[ubi-GFP], FRT42D/CyO females (stocks described in FlyBase)were crossed with w; chn^ECJ1, FRT42D/CyO males. Recombination was induced by heat treatment at 37°C for 30 minutes(Xu and Rubin, 1993). To generate clones of cells overexpressing UAS-chni, males carrying this transgene were crossed with y w FLP122; act-FRT y⁺FRT-Gal4 UAS-GFP / SM6a-TM6b Tb females and recombination was induced by incubation at 37°C for 10 minutes.

Antibody staining was performed as described(Cubas et al., 1991). Primary antibodies were: anti-Sens (Nolo et al.,2000), anti-Sc (Skeath and Carroll, 1991), anti-Ato(Jarman et al., 1995), mAb 22c10 (Zipursky et al., 1984)and anti-β-galactosidase (Cappel). A guinea pig anti-Chn antibody was prepared against a His-tagged Chn fragment corresponding to amino acids 213-417 cloned into the pET28a vector. However, it only allowed clear visualization of the Chn protein under overexpression conditions(Fig. 5N,O). Secondary antibodies were from Jackson and Amersham. In-situ hybridizations to detect chn mRNA were performed as described(González-Crespo and Levine,1993) using an antisense DIG-labeled RNA probe.

A Chn protein fragment (Chn5ZF) that contained the five zinc finger motifs(Asn 297-His 604) was produced in Escherichia coli. This protein was purified with the NTA-system (Qiagen). Six partially overlapping probes of approximately 300 nucleotides each were synthesized, which together covered the entire sequence of the 1.4 kb DNA fragment containing the DC enhancer(García-García et al.,1999). The primers used were: DC1 probe,5′-GAGGAACAAAGAGCAGG-3′ and 5′-TTATAGTCCCCACTG-3′; DC2 probe, 5′-AAACCGCAGCAGTTC-3′ and 5′-GGAATGAGATTGCGG-3′; DC3 probe,5′-AAAAAACCGCCGCTG-3′ and 5′-AACTTTCCCTGCACC-3′; DC4 probe, 5′-ACATATTTCCGGCGC-3′ and 5′-TGTACGACTACAGGC-3′; DC5 probe,5′-TGGTAGGGTAGGATC-3′ and 5′-GACTTATCGTCACGG-3′; DC6 probe, 5′-TTCATTCATCCGGCG-3′ and 5′-GTCGACTTTCGGTTTTTCG-3′. Probes were labeled with[α-³²P]dCTP (3000Ci/mmol) by PCR. The composition of the binding buffer was 50 mmol/l Tris-HCl (pH 7.5), 50 mmol/l KCl, 10% glycerol, 1 mmol/l DTT, 100 μmol/l ZnSO_4, 7.5 μg/ml poly(d(C-I)). The amount of Chn5ZF used was 2 μg for each binding reaction.

The chn locus was first identified in a screen for lethal P elements that affected embryonic development(Kania et al., 1995). The chn insertion (chn^42/18; Fig. 1A), located at chromosomal subdivision 51EF, caused an abnormal morphology of the larval PNS neurons, some of which appeared enlarged, while others, such as those of the lateral chordotonal organs, appeared bunched. In an independent experiment, we mobilized an EP element from a nearby locus and obtained the insertion EPIL6 located near chn^42/18 at the 5′ end of CG11798 (Fig. 1A). The EP elements carry several Gal4 UAS binding sites and can therefore be crossed with Gal4 drivers to ectopically express adjacent genes(Rørth, 1996). Expression of drivers that direct expression of Gal4 in the wing imaginal disk induced formation of additional bristles, mostly macrochaetae(Fig. 1B,C), suggesting that CG11798 is involved in bristle formation. We obtained cDNAs for CG11798 and aligned their sequences with that of the genomic chn DNA. As shown in Fig. 1A, there were two classes of cDNAs, which shared a common short(273 bp) 5′ region, indicating the presence of two types of transcripts resulting from alternative splicing. The first class of cDNAs, probably entirely colinear with the genomic DNA, contained an ORF (630 nt), that putatively encoded a polypeptide with no recognizable similarities to known motifs or proteins. We named this transcript `belinda'(bda), after the classical speechless film character. The second class of cDNAs resulted from spliced transcripts and coincided with the predicted gene CG11798. It encoded a putative 1108 amino acid protein with five C2H2 zinc finger motifs. The zinc finger region showed 35% identity and 55% similarity to the human `Zinc finger protein 462'(Fig. 1A). As shown below, the transcripts encoding the Drosophila zinc finger protein were responsible for chn function in the developing PNS.

When driven by en-Gal4, the EPIL6 insertion induced the overexpression of both chn and bda transcripts in wing imaginal disks (not shown). Hence, we created flies carrying either a UAS-chn or a UAS-bda transgene. Overexpression of UAS-bda using several drivers (MS1096-Gal4, C765-Gal4,ap-Gal4 and MS248-Gal4) did not cause noticeable phenotypic effects (data not shown). By contrast, overexpression of UAS-chn with these and other drivers gave rise to extra bristles. Ubiquitous expression with the C765-Gal4 driver(Gómez-Skarmeta et al.,1996) caused the appearance of many macrochaetae near wild-type bristles, and it increased the density of microchaetae [141±10 microchaetae per female heminotum versus 103±7 in Oregon R controls(averages of 10 heminota) and Fig. 1D]. However, the bristles were always separated by epidermal cells, suggesting that Notch-mediated lateral inhibition(Artavanis-Tsakonas et al.,1999) was still active. With earlier-expressing drivers such as MS248-Gal4 (Cavodeassi et al.,2002), overexpression gave rise to many bristles (see Fig. 4D). In addition, this early expression reduced the size of the heminota and interfered with their dorsal fusion. With the appropriate drivers many extra bristles appeared on other regions of the fly body, including wings (C765-Gal4,MS1096-Gal4 and nub-Gal4), head (MS248-Gal4) and the metathorax (MS1096-Gal4) (data not shown).

Formation of additional bristles by overexpression of chn depended on the presence of the proneural genes ac/sc. Halving the dose of these genes sharply reduced the number of extra macrochaetae induced by the overexpression of chn (compare Fig. 1E with 1D). Removal of both ac and sc completely suppressed bristle formation(Fig. 1F). The genetic interaction between sc and chn was also manifested by the synergism of their overexpression in bristle formation(Fig. 1G-I), which gave rise to many macro- and mesochaetae in ectopic positions, such as the anterior notum(Fig. 1I).

We examined the patterns of expression of chn in embryos and imaginal disks using in-situ hybridization. In early blastoderm stages, the expression of chn was ubiquitous, but before stage 5, chnmRNA disappeared from the poles of the embryo and faint stripes became visible(data not shown). At stage 5, chn mRNA also accumulated in the dorsal region, cephalic furrow and in the presumptive mesoderm(Fig. 2A,B). At stage 11, chn mRNA was found mostly in the mesoderm(Fig. 2C), and in ectodermal patches between the tracheal pits (Fig. 2D,E), where neurons of the PNS appear(Ruiz-Gómez and Ghysen,1993). Older embryos (stage 15, Fig. 2F) showed strong expression, which was mostly restricted to the central nervous system (CNS)and PNS. In the latter case, the pattern suggested that expression occurred in many of the neurons of the ventral, lateral and dorsal clusters of neurons(Fig. 2F,G; compare with Fig. 3A).

In third instar wing disks, expression of chn is observed in rows of cells on either side of the prospective anterior wing margin and in groups of cells that coincided with proneural clusters of ac/sc expression(Fig. 2H). This pattern suggests that chn may be positively regulated by ac/sc. Indeed, expression of chn in proneural clusters was abolished in disks null for ac/sc (In(1)sc^10.1)(Fig. 2I). Moreover,overexpression of a UAS-sc transgene in the posterior compartment of the disk (en-Gal4 driver) induced ectopic expression of chn(Fig. 2J). Note that chn was also expressed independently of ac/sc in certain areas of the disk, such as the postnotum and posterior dorsal proximal wing(Fig. 2H,I, arrows). chn was also expressed in proneural clusters of the leg disks(Fig. 2K) and in the eye/antenna disk (Fig. 2L). In the latter case, the pattern seemed very similar to that of the proneural gene atonal (ato), and included the region of the morphogenetic furrow, and the presumptive cephalic capsule and second antennal segment.

To examine the effects of the removal of chn, we obtained new LOF alleles by generating imprecise excisions of the Pl(2)42/18insertion. One of these, chn^ECJ1, is probably a null, as the excision removed at least part of the promoter region of chn(Fig. 1A), and homozygous embryos lacked the chn mRNA (as detected by in-situ hybridization,not shown) and died as embryos. In keeping with the expression of chnin the cells of the PNS, chn^ECJ1 embryos displayed conspicuous anomalies in PNS cells. These included the absence of many neurons, especially in the dorsal and ventral clusters, and an abnormal morphology of chordotonal lateral neurons, which appeared bunched and lacked the typical apical dendrites (Fig. 3A,B, insets). Individually identifiable neurons such as the v'chn1 and the dbp were generally absent(Fig. 3B). Some of these defects were similar to the phenotype described for the original Pl(2)42/18 insertion (Kania et al., 1995) but were more severe. Ubiquitous expression of UAS-chn in the epidermis (69B-Gal4 driver)(Brand and Perrimon, 1993)largely rescued many of the missing neurons(Fig. 3C) and the morphological defects of the lateral chordotonal neurons(Fig. 3C, inset). However,overexpression of UAS-chn in a wild-type background (using da-Gal4, 69B-Gal4 and 1407-Gal4 drivers) did not appreciably affect the larval PNS (data not shown).

In the adult, the effects of chn^ECJ1 were examined through clonal analysis. Clones of cells that lacked chn often failed to generate macrochaetae (Fig. 4A,B). These were observed to be missing in all notum positions,but scutellar bristles were the most sensitive. However, the penetrance was far from complete: clones including the dorsocentral (n=24) or the posterior postalar (n=8) positions lost approximately 25% of the bristles, while in scutellar clones (n=52) about 45% of the bristles were removed. We also observed, with low frequency, that the bristle shaft was missing, but not the tormogen cell, suggesting a role of chn in formation of the sensory organ. No defects were seen in the pattern of notum microchaetae.

We further examined the effects of the removal of chn in macrochaetae formation with the help of an RNA-interference construct(Sharp, 2001), UAS-chni. A strong (UAS-chni^S) and a weak(UAS-chni^W) expressing line were used. UAS-chni^S clearly antagonized chn function,because it largely rescued the extra bristle and heminota fusion phenotypes of MS248-Gal4; UAS-chn (Fig. 4D,E). No rescue was observed by replacing UAS-chni^S by UAS-lacZ or UAS-GFP, which indicated that the effect was not caused by reduced expression of UAS-chn in the presence of an additional UAS transgene. Moreover,overexpression of UAS-chni^S with MS1096-Gal4sharply decreased accumulation of the endogenous chn mRNA in the wing margin (not shown, see below). These experiments indicated that the UAS-chni^S acted as a LOF allele of chn. With the drivers ap-Gal4 or sca-Gal4 (the latter promotes expression in proneural clusters), UAS-chni^S moderately removed notum macrochaetae (Table 1), while microchaetae were not affected. With the MS248-Gal4 driver, the macrochaetae in the medial part of the notum, dorsocentrals and scutellars,were often missing (Fig. 4C),but, similar to the homozygous chn^ECJ1 clones, in no case was the phenotype fully penetrant.

The effectiveness of both UAS-chni^S and UAS-chni^W in removing macrochaetae was increased when the accumulation of ac/sc in proneural clusters was compromised(Table 1). Thus, flies overexpressing UAS-chni^W with C765-Gal4 or pnr-Gal4 had, respectively, wild-type or almost wild-type phenotypes(Table 1), but a sizable number of macrochaetae were lost in the heterozygous In(1)sc^10.1/+ genetic background. The N allele Ax^M1 has been shown to reduce ac/sc expression in proneural clusters (Martínez-Arias et al., 2002). Hence, in Ax^M1/+ individuals,several notum macrochaetae are missing. UAS-chni^S almost completely eliminated the remaining bristles in Ax^M1/+with the exception of the ANP and APA that were always present(Table 1). The pnrprepattern gene is necessary for the formation of the dorsocentral and scutellar macrochaetae(García-García et al.,1999; Heitzler et al.,1996). In the hypomorphic pnr^V1/pnr-Gal4background, expression of ac/sc is diminished and these bristles are partially removed. In this genetic background, UAS-chni^Walmost completely eliminated all these bristles(Table 1). Interestingly, UAS-chni^W was able to partially suppress the extra macrochaetae that were generated by the expression of UAS-sc driven by C765-Gal4 (Fig. 4F,G): the number of macrochaetae per notum decreased in the dorsocentral area from 12±1.2 to 7.5±1.2 and, in the scutellum,from 18±1.3 to 10.2±1.3. Moreover, macrochaetae that appeared away from the normal macrochaeate-bearing regions, such as the anterior notum,were almost completely eliminated (from 5.4±1.3 to 1±1.2). This suggested that these extra macrochaetae had a stronger requirement for chn than the extant bristles. Taken together, these and the data obtained from the chn^ECJ1 clones indicate that chn is not essential for macrochaetae formation, but that it facilitates the process.

The strong genetic interaction between the LOF conditions for chnand ac/sc, together with the presumed activation of chn by ac/sc (Fig. 2H-J), led us to examine whether chn might in turn stimulate ac/scexpression. We first examined whether the overexpression of chnaffected Sc accumulation in third instar wing disks. In these disks, ac and sc are coexpressed in a stereotyped pattern of well-resolved proneural clusters from which SOPs emerge(Fig. 5A)(Cubas et al., 1991; Skeath and Carroll, 1991). With the MS248-Gal4 driver, UAS-chn promoted strong and generalized expression of sc in most of the domain of expression of the driver, namely, the medial and part of the lateral prospective notum(Fig. 5C). Many SOPs arose from this enlarged region of Sc accumulation, as detected by the Sens marker(Nolo et al., 2000),consistent with the additional macrochaetae that developed on the notum of these flies (Fig. 4D). With the MS1096-Gal4 driver, which is expressed most strongly in the dorsal part of the wing anlage (Capdevila and Guerrero, 1994), there was also ectopic expression of scand emergence of extra SOPs in the wing territory(Fig. 5B). Interestingly,expression of UAS-chn disrupted the characteristic double row expression of sc and sens at the wing margin, suggesting interference with its formation (Fig. 5B). This is also consistent with the presence of small, crumpled adult wings that carry many bristles and other types of sensilla (data not shown). Finally, overexpression of UAS-chn with the ubiquitous wing disk driver C765-Gal4(Gómez-Skarmeta et al.,1996) activated sc but failed to stimulate atonal (Jarman et al.,1993b), a proneural gene which is not a member of the ASC and is normally expressed in a few cells at the presumptive tegula and ventral radius(Fig. 5D). Conversely,overexpression of atonal did not stimulate chn in the wing disk (not shown).

The expression of ac/sc in proneural clusters is controlled by a series of separable enhancer elements in the ASC. Each enhancer is responsible for expression in one or in a few proneural clusters(Gómez-Skarmeta et al.,1995). We thus examined whether the ectopic activation of sc could be mediated by the overexpression of UAS-chn acting upon these enhancers. As shown in Fig. 5E,F, UAS-chn strongly stimulated the activity of a construct in which the lacZ gene was under the control of the ASC L3/TSM enhancer [construct 2.3-lacZ(Culí and Modolell,1998); MS1096-Gal4 driver], which directs expression at the wing vein L3 and the twin sensilla of the wing margin proneural clusters. Similar observations were made with the dorsocentral (DC) enhancer [construct AS1.4DC=DC-lacZ(García-García et al.,1999); C765-Gal4 ubiquitous driver], which promotes expression in the central part of the notum. It also activated expression directed by the ANP enhancer(Gómez-Skarmeta et al.,1995) (data not shown). Since the DC-lacZ construct bears the heterologous hsp70 promoter, these data indicate that the sc endogenous promoter is dispensable for the stimulation by UAS-chn. The Ac and Sc proneural proteins were also not essential for the increased activity of the enhancers, as DC-lacZ expression was strongly increased by Chn in an In(1)sc^10.1 background(Fig. 5I,J). By contrast, the sc SOP-dedicated enhancer (SRV-lacZ construct), which is responsible for the strong accumulation of Sc in SOPs(Culí and Modolell,1998), was only clearly activated by UAS-chn(C734-Gal4 driver) in the presence of ac/sc, and this stimulation occurred in individual cells(Fig. 5N,O). This observation suggests that the upregulation of this enhancer results from the formation of ectopic SOPs by the UAS-chn-induced overexpression of sc,rather than from a direct effect of Chn on the enhancer. Still, the possibility remains that Chn and Sc cooperate in the activation of this enhancer.

UAS-chn upregulated the activity of these enhancers, but it did not lead to a generalized expression of lacZ in all the domains of UAS-chn expression. These data indicate that despite the elevated activation, the enhancers are still dependent on the prepattern factors that define their spatial domains of activity. This fact was verified by the observation that the overstimulation of DC-lacZ was strongly dependent on its prepattern activator, the transcription factor Pnr(Fig. 5K-M). Moreover, 2.3-lacZ, which is active only in the wing pouch, was not stimulated by the overexpression of UAS-chn in the prospective notum(MS248-Gal4), which indicates that the sc promoter present in this construct was not responsive to UAS-chn (data not shown).

Next, we examined in mosaic wing disks whether removal of chnfunction affected expression of sc or enhancer-lacZconstructs in proneural clusters. Homozygous chn^ECJ1 cells generally displayed reduced expression of sc(Fig. 6A) orβ-galactosidase under the control of proneural enhancers(Fig. 6B,C,F,G), when compared with neighboring heterozygous chn^ECJ1/+ cells. Note,however, that the expression was not completely abolished. Similar decreased expression of sc was observed by misexpressing UAS-chni^S in cell clones(Fig. 6D). These effects appear to be cell-autonomous. While SOPs could still emerge from homozygous chn^ECJ1 cells with reduced levels of Sc (data not shown),SOPs were often missing (Fig. 6H), in agreement with the partial suppression of macrochaetae observed within the chn^ECJ1 clones. When both homozygous and heterozygous cells were near a position where an SOP emerged, a heterozygous cell appeared to be preferentially selected(Fig. 6C,E). These findings clearly indicate that chn⁺ was required for proneural proteins to accumulate in proneural clusters at levels sufficient to ensure SOP selection. Moreover, the observation that expression of enhancer-lacZ constructs was reduced in chn^-cells and increased in chn overexpressing cells indicates that the effect of chn⁺ was not due to an enhanced perdurance of the Sc protein, but to the increased transcription of the scgene.

To analyze whether Chn is a direct regulator of the ASC enhancers, we assayed the ability of the Chn protein to bind to the DC enhancer in vitro. A fragment of the Chn protein containing the five zinc fingers was produced in and purified from E. coli. The enhancer DNA was divided into six partially overlapping fragments of approximately 300 bp each, and each of them was assayed in gel retardation experiments(Fig. 7A). Only the fragment that comprised the proximalmost region of the enhancer (fragment DC6) showed binding of the Chn polypeptide (Fig. 7B). Interestingly, the DC6 fragment is included within the PB0.5DC sequence (Fig. 7A)(García-García et al.,1999), the smallest subfragment of AS1.4DC, which still retains enhancer activity. For unclear reasons, the PB0.5DC enhancer only drives expression in the PDC SOP (Fig. 7C)(García-García et al.,1999). Still, misexpression of UAS-chn expands this expression to many cells of the posterior notum(Fig. 7D). This suggests that the binding of Chn protein to the DC6 region of the DC enhancer may prompt its response to Chn in vivo.

We have identified chn, a novel gene that encodes a zinc finger factor that is involved in the development of the PNS of the Drosophila embryo and the adult fly. We have examined in detail the function of chn in the formation of the stereotyped pattern of notum macrochaetae. Complete removal of chn expression led to a relatively mild phenotype; namely, the failure of each notum macrochaetae to develop in 25 to 45% of the flies. Any macrochaeta was subject to loss. This loss was strongly enhanced when, concomitant to the removal of chn, the proneural function of ac/sc was reduced by either halving the doses of the ASC [a condition that normally does not cause the loss of any macrochaeta (García-Bellido,1979)] or by introducing alleles that decreased accumulation of Ac/Sc in proneural clusters. This result suggested a positive interaction between proneural and chn functions in macrochaetae development, an inference that was verified by overexpression experiments. Thus,overexpression of chn gave rise to a large number of extra macrochaetae, an effect that was strongly dependent on the number of doses of the ASC. Reciprocally, the number of extra macrochaetae that arose when overexpressing sc was sharply decreased by compromising chnfunction. In all cases, the extra macrochaetae that were formed upon chn overexpression were not contiguous to one another and epidermal cells were present between them. This indicated that N-mediated lateral inhibition (reviewed by Artavanis-Tsakonas et al., 1995) was still operating and that chn was unlikely to antagonize this process.

The presence of chn mRNA in the proneural clusters of the wing disk is dependent on ac/sc. Moreover, ectopic accumulation of Sc results in ectopic expression of chn. These observations place chn downstream of ac/sc, and suggest a positive, possibly direct, regulation of chn by ac/sc. Consistent with this,two clusters of four and eight E-boxes, putative binding sites for bHLH proteins of the proneural type (reviewed by Bertrand et al., 2002), were found approximately 15 kb upstream of the chn structural sequences and within the first intron of the gene, respectively.

In turn chn stimulates the accumulation of Sc in proneural clusters, as loss of function of chn resulted in decreased accumulation of Sc. However, some Sc still accumulates in the complete absence of Chn, which probably explains why many SOPs and their corresponding macrochaetae developed in its absence. The upregulation of sc by chn is even more manifest by the overexpression of UAS-chn,which causes a strong accumulation of Sc and leads to the formation of large numbers of SOPs and extra macrochaetae. ac is also upregulated by overexpression of chn (L.M.E., unpublished). Although we cannot rule out that Chn may slow the turnover of Sc/Ac and thereby promote their accumulation, our data clearly show that Chn stimulates the transcription of ac/sc. Indeed, the overexpression of chn greatly increases in vivo the expression of the reporter gene lacZ driven by proneural group-specific enhancers of the ASC(Culí and Modolell,1998; García-García et al.,1999; Gómez-Skarmeta et al., 1995) and its removal decreases the expression of these constructs. The stimulation is also observed with enhancer constructs that do not have the endogenous sc promoter (rather, they carry an hsp70 minimal promoter). These data suggest that chn acts mainly on the ASC enhancers, but we cannot rule out at present that the endogenous promoter might additionally favor this effect. However, our results argue against a stimulatory action of Chn directly on the sc and/or ac promoters, since generalized expression of UAS-chn did not lead to widespread expression of the constructs carrying the scpromoter. Moreover, the stimulation was equally observed in the presence or absence of the endogenous ac/sc genes, which indicates that it is not mediated by positive feedback loops of ac/sc on the ASC enhancers, in agreement with previous observations(Gómez-Skarmeta et al.,1995). Considering that the ASC enhancers act in vivo on both the sc and the ac promoters(Cubas et al., 1991; Gómez-Skarmeta et al.,1995; Skeath and Carroll,1991), it was to be expected that Chn would also stimulate ac expression.

Interestingly, Chn not only increased the levels of lacZexpression within the proneural cluster for which the enhancer was specific,but in general it also expanded the expression into a larger area surrounding the proneural cluster, so that more cells were expressing the reporter gene. Perdurance of β-galactosidase should not be responsible for this effect,because when chn was not overexpressed, DC-lacZ directedβ-galactosidase accumulation only in the cells that also expressed sc at the DC cluster(García-García et al.,1999). Moreover, the stimulation by chn seemed to require the presence of at least some of the prepattern factors (reviewed by Ghysen and Dambly-Chaudière,1988; Gómez-Skarmeta et al., 2003) that normally act on the enhancers and drive the expression of ac and sc in proneural clusters, as is the case for Pnr, the prepattern activating factor of the DC cluster(García-García et al.,1999). We propose that excess Chn makes the proneural cluster enhancers responsive to suboptimal concentrations of the prepattern activators that are normally too low to permit activity. Hence, the domains of expression of lacZ are expanded. The dependence of Chn stimulation on different prepattern factors suggests that Chn acts as a coactivator, increasing the effective interaction of prepattern activators with the ac and sc promoters. Moreover, the finding that a fragment of Chn that contains the five Zn-finger motifs of the protein can bind in vitro to a 316 bp fragment of the DC enhancer DNA further suggests that Chn stimulates ac/sc expression by directly binding to ASC proneural cluster-specific enhancers. The possible functional relevance of this binding is reinforced by the fact that the 316 bp fragment is found within a 508 bp segment that possesses residual DC enhancer activity and that Chn is capable of strongly stimulating this activity in vivo.

Chn does not appear to act in vivo as a general stimulator of the enhancer action of proneural genes. The ASC enhancer(s) responsible for expression of ac/sc during microchaetae formation did not require Chn, as judged by the independence of microchaetae density from the activity of chn. Note that downregulation of ac and/or sc normally leads to a strong loss of microchaetae(Ruiz-Gómez and Modolell,1987). By contrast, overexpression of UAS-chn did increase their density, suggesting that the microchaetae enhancer(s) can potentially respond to Chn. chn^ECJ1 clones and UAS-chni did not alter the anterior wing margin bristles. However,overexpression of UAS-chn impaired the expression of sc at the anterior wing margin (Fig. 5A-C), although we favor the idea that this inhibition results from an interference of Chn with the general patterning of the wing, as suggested by the inhibition of sens expression even in the posterior wing margin (Fig. 5A). The lack of an identified ASC wing margin enhancer has prevented a more direct test of these possibilities. We also found that the ASC SOP-specific enhancer(Culí and Modolell,1998) could not be activated in the absence of ac/sc and that the stimulation that we observed occurred in isolated cells, rather than in the majority of cells of the domain of UAS-chn expression. Probably, the stimulation resulted from extra SOPs arising from the overexpression of the endogenous sc gene. Finally, the proneural gene atonal, which is not a member of the ASC(Jarman et al., 1993b), was not affected in the wing or in the eye (L. M. E., unpublished) disks by UAS-chn. We conclude that in the wing disk, Chn is mostly specific for the ASC enhancers that direct ac/sc expression in the proneural clusters of the macrochaetae and other landmark sensilla, such as the twin sensilla of the anterior wing margin (TSM) and the L3 wing vein sensilla campaniformia.

Taken together, our data indicate that chn and ac/sc form a mutually stimulatory loop that enhances accumulation of Ac/Sc in the proneural clusters of the notum macrochaetae(Fig. 7F). These and other previous findings suggest the following consecutive levels of genetic control during SOP selection. The process starts by the deployment of combinations of prepattern factors that trigger the expression of ac/sc in proneural clusters (reviewed by Ghysen and Dambly-Chaudière, 1988; Gómez-Skarmeta et al.,2003) (Fig. 7E). Then, ac/sc activate chn and their stimulatory loop reinforces the expression of ac/sc(Fig. 7F). This allows increasing levels of Ac/Sc to accumulate in the cells of the proneural cluster and the formation of the proneural field, which includes the few cells of the cluster with the highest levels of Ac/Sc(Cubas and Modolell, 1992). The SOP will be selected from one of these cells by the Ac/Sc-mediated activation of sens, which in turn allows the autostimulatory loops of the proneural genes mediated by the SOP-specific enhancers(Culí and Modolell,1998; Jafar-Nejad et al.,2003; Nolo et al.,2000) (Fig. 7G). These enhancers are the targets of two antagonistic signaling systems, both triggered by the accumulation of Ac/Sc. The positive one is mediated by the EGF receptor (Culí et al.,2001) and Sens. The EGFR pathway allows the cells of the proneural cluster to signal positively to each other (lateral cooperation) and helps activate the SOP-specific enhancers, whereas Sens directly activates proneural gene expression in a positive feedback loop when the proneurals reach a certain threshhold in the SOP. Sens and EGFR are in turn antagonized by the negative loop, which is mediated by the Dl/N pathway and the E(spl) proteins and prevents more than one cell from turning on the proneural gene self-stimulation and becoming an SOP (lateral inhibition)(Artavanis-Tsakonas et al.,1995; Culí and Modolell, 1998; Giagtzoglou et al., 2003; Nolo et al.,2000; Heitzler and Simpson,1991; Jafar-Nejad et al.,2003; Simpson,1990; Simpson,1997). Thus, three loops of self-stimulation of ac/scexist: the first is mediated by chn and targets the proneural cluster enhancers; the second is mediated by the EGFR pathway and targets the SOP-specific enhancers; the third is mediated by Sens and also directly targets the SOP-specific enhancers. It is interesting to note that the first and third stimulatory loops are mediated by Zn-finger transcription factors of the C2H2 type with homologs in mammals and other species. The negative loop,mediated by Dl/N and the E(spl) proteins, maintains most cells of the proneural cluster in a non-SOP state, allowing them to differentiate as epidermal cells (Fig. 7E,F). As previously discussed, it is tempting to speculate that these consecutive layers of control facilitate the refinement of the position where SOPs arise within proneural clusters (Culí et al., 2001).

In the embryo, chn is expressed in regions where the neurons of the PNS will arise and later in the developing neural cells. Its removal causes loss of PNS neurons and defects in the morphology of the chordotonal organs, suggesting that chn is required for the proper formation of many or most elements of the PNS. So far, the reported effects of insufficiency of proneural gene function in the embryonic PNS have mostly been the removal of neurons and chordotonal organs, rather than defective morphologies (Dambly-Chaudière and Ghysen, 1987; Huang et al.,2000; Jarman et al.,1993b; Villa-Cuesta et al.,2003). Hence, we like to suggest that in the embryonic PNS chn acts more as a neuronal differentiation gene than a proneural gene activator. In agreement with this suggestion we observed that overexpression of UAS-chn did not modify the embryonic PNS, as detected with the 22c10 antibody. By contrast, overexpression of proneural genes promotes development of extra neurons and chordotonal organs(Huang et al., 2000; Jarman et al., 1993b; Villa-Cuesta et al., 2003). Moreover, loss of function of cousin of atonal (cato) and ase, two genes that can act as neuronal differentiation genes, also causes malformations of the lateral clusters of chordotonal organs(Goulding et al., 2000). We do not know whether the removal of chn may also affect the differentiation of the adult bristles, but the observation that, with low frequency, a shaft can be missing, but not the basal cell, also suggests a role of chn in the differentiation of these SOs. Moreover, the fact that UAS-chni partially suppressed the extra macrochaetae induced by UAS-sc (Fig. 4F,G), a transgene not subjected to chn modulation, may additionally indicate that chn favors macrochaetae formation. However, it should be kept in mind that UAS-sc may promote accumulation of Sc not only through its own expression, but also by the activation of chn, which would in turn stimulate the endogenous ac/sc genes. This latter stimulation should be sensitive to UAS-chni and its inhibition might partially suppress the formation of extra macrochaetae. At present, we cannot decide on these alternatives.

We are grateful to: S. Campuzano, S. Sotillos, J. L. Gomez-Skarmeta, L. Sanchez, R. Barrio, M. Suzanne, N. Azpiazu, E. Sanchez-Herrero, J. F. de Celis, G. Morata and colleagues of J.M.'s laboratory for advice on the work and constructive criticism of the manuscript; to C. Estella and S. Aldaz for providing reagents and stocks. Predoctoral fellowship from Comunidad Autónoma de Madrid to L.M.E. is acknowledged. H.J.B. is an investigator of the Howard Hughes Medical Institute. A grant from Dirección General de Investigación Científica y Técnica (BMC2002-411) to J.M. and an institutional grant from Fundación Ramón Areces to the Centro de Biología Molecular Severo Ochoa are acknowledged.