Drosophila Tufted Is a Gain-of-Function Allele of the Proneural Gene amos
Eric C. Lai


Tufted is a classical Drosophila mutant characterized by a large number of ectopic mechanosensory bristles on the dorsal mesothorax. Unlike other ectopic bristle mutants, Tufted is epistatic to achaete and scute, the proneural genes that normally control the development of these sensory organs. In this report, I present genetic and molecular evidence that Tufted is a gain-of-function allele of the proneural gene amos that ectopically activates mechanosensory neurogenesis. I also systematically examine the ability of the various proneural bHLH proteins to cross-activate each other and find that their ability to do so is in general relatively limited, despite their common ability to induce the formation of mechanosensory bristles. This phenomenon seems instead to be related to their shared ability to activate Asense and Senseless.

ALTHOUGH the nervous system of the fruitfly is quite complex, it is also highly stereotyped. These characteristics make it an ideal experimental system for understanding basic principles of pattern formation. Accordingly, studies of how the Drosophila nervous system is assembled have occupied the collective efforts of hundreds of developmental geneticists over the decades.

The pattern of sensory organs in both the embryo and the adult is prefigured by the spatially patterned expression and activity of the proneural genes, which encode basic helix-loop-helix (bHLH) transcriptional activators (reviewed by Bertrandet al. 2002). Proneural activity confers neural potential upon groups of adjacent cells referred to as proneural clusters (PNCs). Interactions among PNC cells restrict this potential to sensory organ precursor (SOP) cells; non-SOP cells of a PNC usually adopt a nonneuronal fate. This process is commonly referred to as “lateral inhibition” and is mediated by cell-cell signaling via the Notch receptor and the products of the neurogenic genes (reviewed by Artavanis-Tsakonaset al. 1999). Once stably specified, the SOP typically undergoes a fixed series of cell divisions to generate the different cells that make up a mature sensory organ, although nonclonally related cells are recruited into certain types of sensory organs. In the case of the mechanosensory bristles, divisions of the SOP generate five cells: a socket cell and shaft cell (which produce structures that are visible from the exterior) and a glial cell, sheath cell, and neuron (which lie beneath the cuticle; Ghoet al. 1999; Reddy and Rodrigues 1999). The glial cell subsequently undergoes programmed cell death, leaving four cells in the mature mechanosensory organ (Fichelson and Gho 2003).

There are two subclasses of proneural bHLH proteins. The Ato class includes Atonal (Ato) and Absent solomultiple-dendritic (MD) neurons and olfactory sensilla (Amos); Ato controls the development of chordotonal organs, R8 photoreceptors, and a subset of olfactory sensilla (Jarman et al. 1993, 1994; Gupta and Rodrigues 1997) while Amos regulates the development of certain MD and olfactory neurons (Gouldinget al. 2000; Huanget al. 2000). Members of the ASC class are encoded by genes in the ac-sc complex (AS-C; Alonso and Cabrera 1988) and include Achaete (Ac), Scute (Sc), and Lethal of scute (L’sc). Ac and Sc are the proneural proteins for the adult mechanosensory bristles (García-Bellido 1979), while L’sc controls the development of the embryonic central nervous system and musculature (Martin-Bermudoet al. 1991; Carmenaet al. 1995). A fourth member of the AS-C, asense (ase), also encodes a protein with an ASC-class bHLH domain. It is not, however, a true proneural protein as it is expressed only by SOPs and not in PNCs; it is likely involved in sensory organ differentiation (Brandet al. 1993; Dominguez and Campuzano 1993). Another transcription factor expressed in SOPs is the zinc finger protein Senseless (Sens); it is involved in maintenance of proneural gene expression (Noloet al. 2000). Although expression of Ase and Sens is normally restricted to SOPs, both exhibit “proneural” activity in that they induce the formation of ectopic sensory organs when misexpressed.

The mechanosensory bristles that cover the exterior of the fly can be observed in live individuals at low magnification; thus, mutations that affect their distribution are easily identified. Two general classes of mutants display extra bristles. Those that compromise lateral inhibition cause multiple SOPs to emerge from an individual PNC, leading to an increase in bristle density or the presence of tight bristle tufts lacking intervening epidermal cells (i.e., Bearded; Leviten and Posakony 1996). In this situation, extra sensory organs arise from the normal complement of proneural clusters. A second class of mutant displays sensory organs in wholly ectopic positions, although these are always separated by epidermal cells. This results from the generation of ectopic proneural clusters, from which singularized SOPs are chosen. Most of the phenotypically strong mutations in the latter class correspond to lesions in hairy (h) or extramacrochaetae (emc; Bridges and Morgan 1923; Moscoso del Prado and García-Bellido 1984). Both encode negative regulators of the proneural proteins: HAIRY is a bHLH protein that directly represses transcription of ac, while EMC is an HLH-only protein that binds to and inhibits the DNA-binding activity of proneural proteins (Van Doren et al. 1992, 1994; Ohsakoet al. 1994). A rarer set of mutants with ectopic PNCs has been designated Hairy-wing (Hw). Once thought to represent a distinct locus in the AS-C, Hw mutants are actually gain-of-function alleles of ac and sc that ectopically activate neurogenesis (Campuzanoet al. 1986; Balcellset al. 1988).

Another mutant with a dramatic ectopic bristle phenotype is Tufted (Tft; Ritterhoff 1952). Surprisingly, it has been little studied since its discovery 50 years ago. In this report, I characterize the cellular basis of the Tft phenotype and show that it results from the establishment of ectopic proneural domains and selection of extra SOPs. Unexpectedly, Tft neither is dependent upon nor significantly cross-activates expression of Ac and Sc, indicating that it harbors proneural activity for mechanosensory organs. Consistent with this, I present genetic and molecular evidence that Tft is a gain-of-function allele of the proneural gene amos that ectopically activates mechanosensory neurogenesis. Misexpression of any of the proneural proteins can promote the development of mechanosensory organs, and I provide evidence that this is not generally due to promiscuous activation of Ac and Sc, but rather to induction of Ase and Sens.


Drosophila stocks: All mutant alleles and transgenic stocks utilized in this study have been previously described: Tft1/SM-TM6B (Ritterhoff 1952), In(1)sc10-1/FM7 (García-Bellido and Santamaria 1978), isoRoi/CyO (Chanutet al. 2002), ato1/TM6B (Jarmanet al. 1994), UAS-ato (Jarmanet al. 1993), daKX136/SM-TM6B (Caudyet al. 1988), sensE2/TM6B and UAS-sens (Noloet al. 2000), ac 2.2-kb genomic transgene (Van Dorenet al. 1994), dpp40C6-Gal4/TM6C (Staehling-Hamptonet al. 1994), bxMS1096-Gal4 (Capdevila and Guerrero 1994), UAS-amos (Gouldinget al. 2000), UAS-sc (Chienet al. 1996), neurA101-lacZ/TM6B (Bellenet al. 1989), m4-lacZ (Bailey and Posakony 1995), hs-Gal4 (Brand and Perrimon 1993), Brd1/ TM6B (Leviten and Posakony 1996), and f, hs-flp; 101E-Gal4 (de Celis and Bray 1997).

Cytology: Tft1/+ polytene chromosomes displayed a cytologically visible aberration at 36F-37A. The nature of the aberration was analyzed using a tiling set of 5-kb digoxigenin-labeled probes representing 100 kb of DNA from the 36F3-7 region. A contiguous set of probes hybridized to an additional band in the 37A region of the Tft1 chromosome, suggesting that the Tft1 aberration involves a duplication and translocation of material from 36F3-7 to 37A. The proximal limit was not determined, but extends a minimum of 75 kb upstream of amos. Two nonoverlapping probes 0-5 and 5-10 kb downstream of the amos start site both showed variable, but modest, amounts of duplicated signal. This suggests that the structure of this end of the aberration is complex, but terminates in the vicinity of amos.

Immunofluorescence: Imaginal discs were processed for immunofluorescence as described previously (Lai and Rubin 2001). The following dilutions for antibodies were used in this study: rabbit α-Scute (1:200, gift of Hugo Bellen), mouse α-Achaete [1:100, Developmental Hybridoma Studies Bank (DHSB)], rabbit α-Asense (1:2500, gift of Yuh Nung Han), rabbit α-Amos (1:4000, gift of Andrew Jarman), rabbit α-Atonal (1:2000, gift of Andrew Jarman), guinea pig α-Senseless (1:5000, gift of Hugo Bellen), mouse α-Delta (1:100, DHSB), mouse α-E(spl)b323 (1:5, gift of Sarah Bray), mouse α-Hindsight (1:50, DHSB), mouse α-Cut (1:100, DHSB), mouse α-β-galactosidase (1:100, DHSB), rabbit α-β-galactosidase (1:5000, Cappel), and rabbit α-Neur (1:400; Laiet al. 2001).


Phenotypic and cellular characterization of Tft: The Tft locus is defined by a single, viable mutant (Tft1) that displays a large number of ectopic mechanosensory bristles, particularly in the postalar, dorsocentral, and scutellar regions of the notum (Figure 1, A and B) (Arnheim 1967; Ritterhoff 1952). I analyzed the expression of several markers of the SOP fate in this mutant, including Hindsight (Hnt; Pickupet al. 2002), Senseless (Sens), neurA101-lacZ (A101), and Asense (Ase). I observed extra cells positive for each marker in the presumptive posterior notal region of Tft/+ wing imaginal discs (Figure 2, A-H, arrows), demonstrating that the Tft phenotype is due to ectopic adoption of the SOP fate. Ectopic SOPs appear to subsequently develop into normal sensory organs in Tft, since ectopic mechanosensory organs contain both sockets and shafts and have neurons that are functionally connected to the central nervous system (Ghysen and Richelle 1979).

Ectopic bristle phenotypes can generally be classified according to whether they arise from ectopic proneural clusters or reflect a failure of lateral inhibition. The term bristle “tufting” is popularly used to refer specifically to a failure of lateral inhibition. Indeed, the presence of closely spaced or even adjacent bristles in Tft flies (Figure 1B) and the determination of SOPs adjacent to each other in Tft wing imaginal discs (Figure 2, E-H) together suggest a defect in lateral inhibition. However, the number of ectopic bristles in the Tft-affected region was significantly increased in Tft1/+; Brd1/+ double heterozygotes (Figure 1E), indicating that Tft bristles are still sensitive to lateral inhibition. In addition, many Tft bristles were seen at clearly ectopic locations, including the anterior-central portion of the scutellum (Figure 1, A and B) and the metathoracic notum (not shown). This suggested the existence of ectopic proneural domains, which are not characteristic of neurogenic mutants.

Figure 1.

—The Tft1 adult phenotype. Shown are scanning electron micrographs (SEMs) of thoraces from adult males of the following genotypes: (A) Canton-S, showing the wild-type pattern of mechanosensory bristles. Inset shows an anterior-central portion of the scutellum that normally lacks sensory organs. (B) Tft1/+, displaying many ectopic macrochaetae and microchaetae; inset shows microchaetae present on the scutellum. (C) sc10-1/Y fly is essentially devoid of mechanosensory organs. (D) sc10-1/Y; Tft1/+ shows ectopic sensory organs in the Tft-affected region. (E) Tft1/+; Brd1/+ shows enhancement of the Tft phenotype; Brd1 heterozygotes display only a few extra macrochaetae in the Tft-sensitive region (Leviten and Posakony 1996). (F) Tft1/dakx136 shows suppression of the Tft phenotype.

I assessed the distribution of proneural clusters using a number of additional markers and observed both elevated and ectopic activity of E(spl)m4-lacZ (as marked by β-galactosidase; Figure 3, A and E, arrows) and expression of E(spl)bHLH proteins (as marked by the MAb323 antibody; Figure 3, B and F, arrows), indicating the presence of ectopic proneural clusters. Notably, these results also indicate that the Tft phenotype is not due to a failure to activate components of lateral inhibition. Surprisingly, I did not observe comparable ectopic expression of Sc and Ac (Figure 3, C, D, G, and H, arrows), the proneural proteins for the adult peripheral nervous system (PNS). Doubly stained preparations showed that ectopic SOPs (as marked by Sens) in the Tft background were not generally associated with corresponding proneural clusters of Ac expression, although Ac could sometimes be observed in ectopic SOPs (Figure 3, I and J). This contrasts with what has been shown for other ectopic bristle mutants such as hairy and extramacrochaetae, which are associated with ectopic clusters of proneural expression and/or activity (Skeath and Carroll 1991; Van Doren et al. 1992, 1994). In summary, Tft is an unusual extra-bristle mutant: It exhibits both ectopic proneural domains and some defect in lateral inhibition, and ectopic neurogenesis is not generally accompanied by corresponding Ac/Sc expression.

Tufted harbors proneural activity for mechanosensory organs: Simultaneous inactivation of ac and sc, the proneural genes for the adult PNS, results in a nearly completely bald fly lacking most mechanosensory organs (sc10-1/Y, Figure 1C). The bald phenotype of sc10-1/Y is epistatic to that of most other ectopic bristle mutants, indicating the fundamental role for these genes in establishing adult peripheral neurogenesis. In contrast, the ectopic bristle phenotype of Tft was epistatic to sc10-1/Y (Figure 1D); similar findings have been noted previously (A. Garcia-Bellido, personal communication cited in Ghysen and Richelle 1979). The ability of Tft to bypass the requirement for Ac/Sc demonstrates that Tft harbors an independent proneural activity. In accord with this, ectopic expression of Sens and Hnt was specifically maintained in the Tft-affected region of sc10-1/Y; Tft/+ wing discs (Figure 2, I-L, arrows). Since neurogenic genes do not exhibit proneural activity, the name “Tufted” is somewhat of a misnomer.

Figure 2.

Tft causes ectopic adoption of the SOP fate in an ac/sc-independent manner. Shown are wing imaginal discs from Canton-S (A-D), Tft1/+ (E-H), sc10-1/Y (I and J), and sc10-1/Y; Tft1/+ (K and L). Ectopic cells express a broad range of markers of the SOP fate in the presumptive posterior notum and scutellum of Tft wing discs (arrows), including Hindsight (Hnt, A and E), Senseless (Sens, B and F), neurA101-lacZ stained for β-galactosidase (A101, C and G), and Asense (Ase, D and H). Ectopic SOPs develop independently of the normal proneural genes for the PNS. sc10-1/Y individuals lack most SOPs (compare brackets in B and J), with the exception of ato-dependent positions (e.g., I, asterisk); nonsensory aspects of Hnt expression [I, tracheal tubes (T)] or Sens expression [J, wing margin (WM)] are unaffected. In this background, Tft still generates ectopic SOPs (I-L, arrows), even though most other SOPs are still absent (L, bracket).

I tested Tft for genetic interactions with other loci high in the regulatory hierarchy for peripheral neurogenesis. Genetic interactions were not observed with either ato or sens, nor was Tft enhanced by increasing ac dosage using an ac genomic transgene (data not shown). However, Tft was partially suppressed by removal of one copy of daughterless (da; Figure 1F), which encodes a bHLH heterodimeric partner for proneural bHLH proteins. In addition, Tft was previously reported to be suppressed by Df(1)260-1, a deficiency of the entire AS-C (A. Garcia-Bellido, personal communication cited in Campuzanoet al. 1985). These interactions suggest that the Tft phenotype might be due to altered bHLH activity.

Tft is associated with an aberration at 36F-37A that results in misexpression of amos: Tft was previously mapped to ∼37A (FlyBase 1998) and shown to be reverted by deficiencies of this cytological region (Wrightet al. 1976), a genetic property consistent with it being a gain-of-function allele. Examination of polytene chromosomes revealed a complex aberration involving a duplication and translocation of sequences at 36F3-7 to 37A (Figure 4, A-D, asterisks; see also materials and methods and the Figure 4 legend for details of this analysis). I noted that the proneural bHLH-encoding gene amos is located at one end of the aberration (Figure 4A); amos is involved in the development of embryonic multiple-dendritic neurons and adult olfactory sensilla (Gouldinget al. 2000; Huanget al. 2000). Since directed misexpression of Amos results in the ectopic production of several types of sensilla, including mechanosensory organs, it was conceivable that Tft is due to misexpression of Amos.

I tested this hypothesis by staining wild-type and Tft tissue for Amos, which is not normally expressed in the wing disc (Figure 4E). Ectopic Amos was indeed observed in the presumptive posterior notal region of the Tft wing disc (Figure 4F) as well as at the base of the haltere disc (data not shown). Consistent with the genetics of Tft, the domain of ectopic Amos was independent of ac/sc (Figure 4G) and included the precise region from which ectopic SOPs are determined in this mutant (Figure 4, H-J).

The findings that Tft specifically misexpresses Amos and interacts genetically with da, which encodes a known heterodimeric partner of Amos (Gouldinget al. 2000; Huanget al. 2000), suggested that Amos is a primary contributor to the Tft phenotype. However, the Tft duplication extends a minumum of 75 kb upstream of amos and affects several additional genes (Figure 4A). In addition, previous reports have differed on the efficacy with which ectopic Amos induces the formation of mechanosensory organs (Gouldinget al. 2000; Huanget al. 2000). I therefore performed additional misexpression experiments to further substantiate the hypothesis that Tft is an allele of amos.

Conditional misexpression of amos efficiently initiates peripheral neurogenesis and induces mechanosensory organ formation: Although prolonged misexpression of amos using drivers such as dpp-Gal4 and bxMS1096-Gal4 resulted in pupal lethality, I was able to characterize their disc phenotypes in detail using the PNC and SOP markers described earlier. Misexpression of Amos with either driver resulted in massive ectopic expression of proneural cluster markers such as Delta (Dl), E(spl)m4-lacZ, and E(spl)bHLHs (Figure 5, A-E, and data not shown); of SOP markers such as Hnt, Sens, Ase, and Neur (Figure 5, F-I, and data not shown); and also led to a significant increase in disc size. Induction of Cut by Amos (Figure 5J, arrow) served as an additional measure of the identity of many ectopic SOPs as precursors for external sensory organs (compare with wild type, inset to J) and contrasted with the activity of the related bHLH Ato, which instead represses Cut (Jarman and Ahmed 1998). Thus, ectopic Amos efficiently establishes new proneural domains and strongly induces external peripheral neurogenesis. Notably, the levels of ectopically produced proteins were generally far greater than those of the corresponding endogenous proteins, suggesting that the neuronal program was “super-activated” by Amos. In spite of this, Sc was only mildly misexpressed in response to Amos (Figure 7G), and ectopic Ac was observed essentially only in the presumptive notum posterior (Figure 7J). In fact, some aspects of Ac expression were somewhat suppressed by Amos. Taken together with the observations that the Tft phenotype does not involve Ac or Sc, I conclude that the neuronal program initiated by ectopic Amos is largely independent of the normal proneural genes for mechanosensory neurogenesis.

Figure 3.

—Upregulation and misexpression of downstream proneural cluster markers in Tft discs. Shown are wing imaginal discs from wild type (A-D) and Tft1/+ (E-J) stained for β-galactosidase (A and E), E(spl)bHLHs stained with Mab323 (B and F), Scute (Sc; C and G), Achaete (Ac; D and H), and Sens + Ac (I and J). Ectopic and increased levels of E(spl)m4 promoter activity (A and E, arrows) and E(spl)bHLH proteins (B and F, arrows) are observed in Tft tissue, whereas Sc (C and G) and Ac (D and H) expression is only mildly affected. Some ectopic cells that adopt a high level of Ac expression are seen in well-stained preparations. (I and J) Higher magnification view of the presumptive posterior notum of a Tft1/+ disc; double staining demonstrates that ectopic SOPs (as marked by Sens; I) are not associated with ectopic clusters of Ac expression, although ectopic Ac is seen in some SOPs (J).

Unexpectedly, commitment to the SOP fate was not generally coincident with expression of Amos on a cell-by-cell basis, in spite of the ease with which Amos induced SOP-specific gene expression. I observed that both Hnt (Figure 5, K-M) and neurA101-lacZ (Figure 5, N-P) were expressed at low levels or not at all by Amos-expressing cells, while cells that accumulated high levels of these markers instead had low levels of or lacked Amos. The same observation applied to Tft tissue as well: Hnt-positive cells in the region displaying ectopic neurogenesis often did not express Amos (Figure 4, H-J). I attempted to assess the autonomy of clones of Amos-misexpressing cells using a FLP-out Gal4 strategy, but this typically resulted in bald patches in the adult, possibly due to toxicity of high and prolonged expression of Amos. However, in the small number of cases where ectopic bristles were formed, they were always Amos+, indicating that induction of sense organs by Amos is likely autonomous (Figure 6C). Consistent with this, Tft1 was likewise previously determined to be cell autonomous in tissue mosaics (Arnheim 1967). Therefore, the inverse relationship between expression of Amos and SOP markers appears to represent negative regulation of Amos in SOPs.

dppGal4>UAS-amos pharate adults were occasionally recovered when they were reared at 18°; these individuals displayed a large number of ectopic mechanosensory organs (Figure 6, A and B). In fact, the ability of UAS-amos to generate ectopic mechanosensory organs with this driver was greater than that of other proneural genes, including UAS-ac, UAS-sc, and UAS-ato; only UAS-sens was on a par with UAS-amos in this regard (data not shown). The lethality of Amos misexpression was reduced by temporal restriction of expression using hs-Gal4 and a 6-min heat shock at 38°. As described previously, these animals displayed a variety of ectopic sense organs (Huanget al. 2000). However, I observed predominant induction of mechanosensory organs and campaniform sensilla (Figure 6, D and E). Thus, misexpression of Amos efficiently induced the development of mechanosensory organs, which is consistent with the Tft phenotype.

Figure 4.

Tft1 is associated with a duplication and translocation in the 36F-37A region that results in misexpression of the proneural gene amos. (A) Physical map of 100 kb from the 36F3-7 region used for cytological analysis of Tft1/+ polytene chromosomes; the map was modified from Gadfly output (Mungallet al. 2003). Selected hybridizations using the probes labeled “B,” “C,” and “D” are shown in B-D, respectively; the positions of 36E (arrowhead) and 37A (arrow) on the wild-type homolog are indicated. DNA from the region shaded yellow hybridized to an additional band in the vicinity of 37A, indicating that the Tft aberration involves a duplication and translocation of sequences from 36F3-7 to 37A. This is most clearly observed in exceptional chromosome figures where the homologs have separated (B); a second site of hybridization is seen on one of the two homologs (asterisks). The proximal limit of the aberration was not determined and lies to the left of the sequence analyzed here. The distal limit terminates in the vicinity of amos (in red), but its structure is complex. Both probe C (including the amos transcription unit) and a probe including the next distal 5 kb of sequence (not shown) showed modest evidence of duplication [C, asterisks designate full hybridization; (*) designates partial hybridization near 37A]. That adjacent but nonoverlapping probes showed this behavior suggests that the distal limit of the duplication does not break cleanly, but has one or more additional anomalies associated with it. The distal limit is therefore shaded in fading yellow. Probe D is distal to the duplicated region and hybridizes to a single band on each homolog (D, asterisks); aberrant pairing of the homologs at 37A is still observed. Amos is not expressed in the wild-type wing disc (E) but is ectopically expressed in Tft tissue in the presumptive posterior notal region (F, bracket) in an ac/sc-independent manner (G, bracket). Ectopic SOPs (as marked by Hnt) develop from the Amos-misexpressing region of Tft discs (H-J).

Limited cross-activating potential of proneural proteins: Most proneural proteins display a significant amount of promiscuous activity when misexpressed, including a common ability to promote the development of mechanosensory organs. To assess whether this was generally attributable to cross-activation of proneural gene expression, I systematically evaluated the ability of proneural proteins (Ac, Sc, Ato, Amos, and Sens) to activate one another when misexpressed using dpp-Gal4 and appropriate UAS constructs. A subset of these data is shown in Figure 7; the results for Ato and Ac are not shown since their misexpression resulted in only very mild cross-activation, at best, of any of these markers when assayed at the third instar.

As noted before, misexpression of Amos only mildly activates Sc (Figure 7G) and Ac (Figure 7J), even though it strongly induced Ase (Figure 7M) and Sens (Figure 7P). Interestingly, Amos also strongly induced Ato, although primarily only in the wing pouch region (Figure 7D, bracket). In contrast, neither Sc nor Sens detectably activated either Amos or Ato (Figure 7, B, C, E, and F). Only Sens induced an appreciable amount of ectopic Sc (Noloet al. 2000) and Ac (Figure 7, I and L), although this was mostly restricted to the dorsal wing pouch and notal regions of the disc. Since ectopic Sc did not generally induce Ac (Figure 7K; Gomez-Skarmetaet al. 1995), this might reflect independent cross-activation of Sc and Ac by Sens. Finally, all three of these proneural proteins could ectopically activate Ase and Sens (Figure 7, M-R), with their rank order of effectiveness being Amos > Sens > Sc. This correlated directly with their ability to activate neurA101-lacZ (Figure 7, S-U). Bearing in mind that one cannot infer direct regulatory relationships from these experiments, it appears that the ability of all of these proneural proteins to activate the mechanosensory organ is probably not due to activation of Ac/Sc, the normal proneural bHLHs for this process. Rather, it is correlated with their common ability to induce Ase and Sens.

Figure 5.

—Directed misexpression of Amos efficiently generates ectopic proneural clusters and SOPs for external sensory organs. Shown are discs from dppGal4>UAS-amos individuals, with the exception of B, which is wild type (wt). In general, the Amos-misexpressing discs are shown at a slightly lower magnification relative to the wild-type disc due to their overgrowth phenotype. Misexpression of Amos in a stripe along the anterior-posterior border (A) results in a similar stripe of ectopic proneural clusters, as marked by Delta (Dl, B and C), m4-lacZ (stained for β-galactosidase, D) and E(spl)bHLHs (E), as well as strong overcommitment to the SOP fate, as marked by Hnt (F), Sens (G), Ase (H), Neur (I), and Cut (J; dR, dorsal radius). Inset to J shows a wild-type dR stained for Cut for comparison; only a small number of cells are normally labeled. Wild-type expression patterns for all other markers are shown in Figures 2 and 3. (K-P) Amos is downregulated following adoption of the SOP fate. The same disc is doubly stained in A and F; the boxed region is shown at higher magnification in K-M. Note that a higher level of Hnt (green) is found in cells that do not express Amos (red). (N-P) Cells that strongly express Amos (red) express low levels of β-galactosidase (green) in the neurA101-lacZ background, while the cells that express the highest levels of β-galactosidase do not express Amos.


amos activates ectopic mechanosensory neurogenesis in Tft: Although the normal function of amos is to initiate the development of certain embryonic multiple-dendritic neurons and adult olfactory sensilla (Gouldinget al. 2000; Huanget al. 2000), several lines of evidence suggest that it is also responsible for generating the dominant Tft phenotype, which consists of a large number of ectopic adult mechanosensory organs.

Figure 6.

—Adult phenotypes caused by misexpression of Amos. SEMs of (A) wild-type and (B) dpp-Gal4>UAS-amos nota are shown; many ectopic macrochaetae are seen in the latter. They generally die as pupae but occasionally survive to the late pharate adult stage; cuticle deformation is due to its dissection from the pupal case. (C) Phenotype of a small clone of Amos-expressing cells that have been marked with forked. Most such clones show only a bald patch; however, exceptional clones also show a tight cluster of forked bristles, suggesting that Amos induces sensory organs cell-autonomously. No instances of clusters of forked+ bristles were ever observed. (D) A proximal-central portion of a wild-type wing. The wing blade proper is devoid of mechanosensory bristles and has only a small number of campaniform sensilla associated with some of the veins (arrow). (E) hsGal4>UAS-amos fly that was exposed to a 6-min heat shock at the white prepupal stage; the wing is covered with several hundred ectopic mechanosensory and campaniform sensilla.

First, Tft maps to the same cytological location as amos and is associated with a chromosomal duplication and translocation that affects amos. Second, Tft mutants ectopically express Amos in precisely the same region from which ectopic SOPs arise in this mutant. Third, Tft is sensitive to the dosage of da, which encodes an obligate bHLH cofactor for proneural proteins such as Amos. Consistent with this, da similarly suppresses Roi (caused by misexpression of amos; Chanutet al. 2002) and enhances the phenotype of amos deficiencies (Gouldinget al. 2000; Huanget al. 2000). Fourth, deliberate misexpression of Amos phenocopies Tft and very effectively generates ectopic mechanosensory organs (in addition to other types of sense organs). Although I cannot formally exclude the contribution of other genes affected by the aberration at 36F-37A, the collected observations strongly suggest that the Tft phenotype can be satisfactorily accounted for by the ectopic expression of Amos. It should be noted that Modolell and colleagues have come to similar conclusions regarding Tft and have also shown that Tft revertants no longer misexpress Amos and/or are hypomorphic for amos (Villa-Cuestaet al. 2003, accompanying article in this issue). Taken together, these observations strongly indicate that Tft is a gain-of-function allele of amos.

The past year has witnessed not only the simultaneous and independent characterization of Tft (this work and Villa-Cuestaet al. 2003), but also the realization that the classical mutant Rough eye (Roi) is likewise a gain-of-function mutant of amos (Chanutet al. 2002). Therefore, this relatively recently identified gene, whose original characterization was also carried out simultaneously and independently by two laboratories (Gouldinget al. 2000; Huanget al. 2000), is affected by two completely distinct gain-of-function mutants. An even more stunning fact is that both Tft and Roi were first described in the same volume of the Drosophila Information Service in 1952 (Ives 1952; Ritterhoff 1952). The history of amos seems thus to be dominated by coincidences.

Unusual features of mechanosensory neurogenesis induced by amos: A curious feature of ectopic peripheral neurogenesis induced by Tft or UAS-amos is that it very minimally involves Ac and Sc, the endogenous proneural proteins for this process. Tft is not suppressed by complete inactivation of these proneural genes and is not modified by an increase in ac dosage. In addition, Sc and Ac are minimally misexpressed in Tft or in directed Amos misexpression experiments, even though all other PNC and SOP markers tested are strongly induced under these conditions. The failure of Amos to induce Sc or Ac is especially surprising considering the fact that Sens is very strongly induced by Amos, and Sens can ectopically induce Sc and Ac (although only in a subset of disc cells). A possible explanation for this paradox is that the high levels of E(spl)bHLH proteins induced by Amos are responsible for repressing ac and sc (Van Dorenet al. 1994; Jimenez and Ish-Horowicz 1997), although it is also the case that Sens can induce E(spl) expression (Noloet al. 2000).

Interestingly, expression of SOP markers such as Hnt and neurA101-lacZ was often inversely correlated with that of Amos on a cell-by-cell level, even though Amos very strongly induces their expression. Since the effects of Tft (Arnheim 1967) and Amos misexpression are autonomous, this probably does not represent nonautonomous induction or recruitment of SOPs, but rather negative regulation of Amos. This control seems unlikely to reside at the transcriptional level in our experiments involving the Gal4/UAS system, suggesting that Amos protein might be unstable in SOPs. amos mRNA is also apparently rapidly lost from olfactory SOPs (Gouldinget al. 2000), suggesting that amos is negatively regulated at multiple levels shortly after commitment to the SOP fate. This behavior seems to run counter to that of Ac and Sc, which accumulate to elevated levels in SOPs, but may be paralleled by Ato, which is downregulated in maturing chordotonal SOPs (zur Lage and Jarman 1999). More detailed studies are required to determine if there are distinct functional consequences of the apparently different regulation of Ato- and ASC-class proteins in SOPs, or if this reflects some trivial difference in the timing of the differentiation of these different SOPs.

Figure 7.

—Cross-regulatory capabilities of selected proteins with proneural activity. All discs contain dpp-Gal4 and UAS-amos (top row), UAS-sc (middle row), or UAS-sens (bottom row). For reference, the misexpressed protein of choice is visualized in A (Amos), H (Sc), and R (Sens); the wild-type expression patterns of most of these proteins are shown in Figures 2, 3, 4. Ectopic Amos activates Ato in the wing pouch region (D, bracket), very mildly activates Sc (G) and Ac (J), and leads to strong misexpression of Ase (M) and Sens (P). By contrast, neither Sc nor Sens activates Amos (B and C) or Ato (E and F) in the wing disc. Arrows in E and F point to normal endogenous Ato expression in the ventral radius (vR); these discs are identical to wild type with respect to expression of Ato (not shown). Sc negligibly activates Ac (K), while Sens activates both Sc (I) and Ac (L), although only in a subset of Sens-expressing cells. Both Sc and Sens can modestly activate Ase (N and O), and Sc misexpression also results in some ectopic Sens (Q). The proneural strength of these proteins can be rank ordered as Amos > Sens > Sc on the basis of their ability to activate neurA101-lacZ (S-U). This correlates directly with their ability to activate Ase and Sens (M-R) and with the strength of their corresponding adult phenotypes (not shown). D and J, E and K, and I and L are from doubly stained preparations; M and P are the same as Figure 5, G and H.

The ability of Tft/amos to induce closely spaced or even adjacent SOPs and sensory organs suggests that it is able to at least partially overcome or bypass lateral inhibition. This is not simply due to disconnecting a proneural gene from its normal transcriptional control, at least in the case of the Gal4-UAS experiments, since misexpression of Ac or Sc by similar means results in ectopic, but spaced bristles. It is also not a consequence of a failure to activate lateral inhibition, since E(spl) bHLH expression is strongly induced by Amos. It may simply be the case that Amos’ unusually potent proneural activity overwhelms or is not very sensitive to lateral inhibition. Another possibility is that induction of exceptionally high levels of E(spl)m4 (and potentially other Brd family proteins) by Amos might interfere with lateral inhibition, an explanation that might underlie the strong genetic interaction between Tft and Brd. Deliberate misexpression of Brd family genes is known to compromise lateral inhibition (Levitenet al. 1997; Apidianakiset al. 1999; Lai et al. 2000a,b).

Another explanation might lie in the difference between the types of sensory organs normally controlled by ASC and Ato-class proneural proteins. While single SOPs for mechanosensory organs are chosen from individual PNCs of ac- and sc-expressing cells, large numbers of SOPs for chordotonal and olfactory sensilla are instead continuously selected from individual zones of ato- or amos-expressing cells (zur Lage and Jarman 1999; Gouldinget al. 2000). It is conceivable that Ato-class proneural proteins possess an inherent ability to induce the formation of closely spaced sensory organs. Signaling via the epidermal growth factor receptor (EGFR) antagonizes N signaling during SOP determination, and the development of clustered chordotonal organs not only relies upon EGFR signaling but also is strongly correlated with localized expression of rhomboid and presence of dp-ERK (zur Lage and Jarman 1999; Culiet al. 2001). It will be interesting to determine if settings of Amos activity are similarly associated with EGFR signaling and if any mechanistic links can be established between expression of Ato-class proteins and activation of rhomboid transcription and/or phosphorylation of MAP kinase.

Specificity of proneural activity: When misexpressed, proneural proteins often induce the ectopic differentiation of sensory structures whose development they do not normally control. For example, Sc can promote many aspects of eye development in the absence of Ato; Sc and Ato can weakly induce the differentiation of MD neurons; and Amos can promote the differentiation of chordotonal organs. Notably, all proneural proteins have the ability to promote the formation of mechanosensory organs (Jarmanet al. 1993; Hinzet al. 1994; Gouldinget al. 2000; Huanget al. 2000; Sunet al. 2000).

Several mutually compatible explanations for their common ability to induce mechanosensory neurogenesis have been put forth. First, experiments with Da suggested that mechanosensory organs might represent a “default” output for neurogenesis. Since Da is the heterodimeric partner for all proneural bHLH proteins, it is not expected to exhibit a subtype specificity. Nevertheless, misexpression of Da induces only the development of external sensory organs (Jarman and Ahmed 1998). Second, most proneural proteins are known to exhibit a phase of transcriptional auto-activation (Martinez and Modolell 1991; Van Dorenet al. 1992; Culi and Modolell 1998; Sunet al. 1998). Although the binding specificities of the proneural proteins are distinct, particularly between the ASC and Ato subclasses, high levels of ectopic proteins could conceivably result in some direct cross-activation of ac/sc. Once initiated, auto-activation by Ac and/or Sc might then be responsible for subsequent mechanosensory organ formation. Third, it is known that the transcription factor Sens, which itself has inherent proneural activity, is downstream of both subfamilies of proneural proteins (Noloet al. 2000; Frankfortet al. 2001). Thus, it may be that the activation of a common downstream target by different proneural proteins could explain how they can all activate mechanosensory neurogenesis.

My data do not speak to the first of these explanations. However, they do suggest that relatively little cross-activation occurs at the level of proneural bHLH gene expression. Although Amos is perhaps the strongest inducer of mechanosensory organs among proneural proteins it does not do so through induction of intermediary proneural clusters of Ac/Sc. Even Sc fails to significantly cross-activate Ac (Gomez-Skarmetaet al. 1995), although Sc and Ac have largely indistinguishable DNA-binding properties in vitro (Singsonet al. 1994). This is not to say that proneural proteins cannot influence each other’s expression, since ectopic Amos can strongly activate Ato in the pouch region of the wing disc and in the vicinity of the morphogenetic furrow of the developing eye disc (Chanutet al. 2002). This particular ability might reflect their related DNA-binding domains; however, it should be noted that Ato does not reciprocally activate Amos (data not shown). Ac, Sc, and Sens also fail to activate either Amos or Ato. So in general, there is limited promiscuity in cross-activation of proneural bHLH proteins.

A common activity of Ato-class and ASC-class bHLH proteins is instead their ability to induce Ase and Sens expression; Sens also induces Ase. As misexpression of either Ase or Sens suffices to initiate mechanosensory organ development, their activation may be key to promiscuous induction of mechanosensory organs. In principle, initiation of an Ase/Da-Sens feedback loop might be responsible for triggering a mechanosensory-type developmental program. Consistent with this scenario, ase is required for ectopic neurogenesis in Tft although ac/sc are not (A. Garcia-Bellido, personal communication cited in Campuzanoet al. 1985; Villa-Cuestaet al. 2003), and Sens is strictly required for bristle formation by ectopic Sc (Noloet al. 2000). The precise contribution of ase to mechanosensory neurogenesis remains in question though, as it is not normally required for the development of most bristles. In addition, both L’sc and Sc can generate ectopic bristles in cells that carry a deletion of the AS-C (and thus lack ase; Hinzet al. 1994; Culi and Modolell 1998). It is conceivable that proneural proteins might generically substitute for ase in these experimental conditions.

Dominant gain-of-function alleles in Drosophila: Although they arise infrequently, dominant gain-of-function alleles can produce dramatic phenotypes that are easily identified, even in the course of unrelated studies. This explains why they are generally among the oldest Drosophila mutants known (Lindsley and Zimm 1992). Although in some cases these mutants have been useful reagents for studying various developmental processes, in other cases, their effects are neomorphic and not informative of normal development. Perhaps for this reason, dominant gain-of-function mutants have generally been treated with suspicion or even de facto discrimination by Drosophila geneticists. Indeed, many “old” dominant mutants have escaped the attention of Drosophila developmental biologists, even though they often affect processes (such as wing, eye, and bristle development) that are otherwise the subject of intense scrutiny. This has remained true in spite of the fact that misexpression studies and systematic gain-of-function genetic screens have become de rigeur since the inception of the versatile Gal4/UAS system in the past decade (Brand and Perrimon 1993; Rørth 1996).

Only very recent years have witnessed the molecular characterization of a number of classical dominant gain-of-function Drosophila mutants, including Rough eye (Chanutet al. 2002), Glazed (Brunneret al. 1999), Lyra (Noloet al. 2001), Drop (Mozer 2001), Beadex (Milánet al. 1998; Shoreshet al. 1998; Zenget al. 1998), and Scutoid (Fuseet al. 1999). In these examples, the mutant phenotype appears to be due to the misexpression of an important regulatory molecule (including transcription factors and a morphogen). Thus, while these mutations are all neomorphic, they are nonetheless useful in that they identify “interesting” genes. Many other commonly utilized dominant mutations—Curly, Tubby, Additional Veins, Blunt short bristle, and Plexate, for example—remain to be characterized. A cursory examination of the FlyBase archive reveals >50 other uncharacterized adult-dominant Drosophila mutants for which mutant stocks are publicly available; about one-half of these mutants have been extant for nearly 50 years or more (FlyBase 1998). Undoubtedly, this collection of mutants, once properly analyzed, will be found to affect many other interesting genes as well.

The history of amos is particularly instructive in this context. Its recent discovery relied upon molecular approaches (degenerate PCR and two-hybrid screening) and specific genetic lesions in amos have yet to be described. Nevertheless, the existence of an amos-like function had been genetically inferred for many years, since certain neurons persist in embryos mutant for both the AS-C and ato, and much of the olfactory system develops independently of these proneural genes. Since the development of all of these neurons is still sensitive to manipulation of Da and/or EMC levels, this suggested the existence of an additional proneural bHLH. Loss of amos function produces phenotypes complementary to AS-C; ato mutants, suggesting it is the “missing” proneural gene (Gouldinget al. 2000; Huanget al. 2000). Had Roi and/or Tft been studied earlier, amos might have been discovered long ago. In this author’s opinion, the very history of amos justifies the study of other long-neglected dominant mutants.


I thank Francoise Chanut, Adina Bailey, and Julia Serano for useful discussions of this work; Juan Modolell for communications regarding Tft prior to publication; and especially Todd Laverty for performing in situ hybridizations to polytene chromosomes. I also acknowledge the following for generous gifts of antibodies and fly stocks: Andrew Jarman, Yuh Nung Jan, Saray Bray, Hugo Bellen, Francoise Chanut, James Posakony, Jose de Celis, the Bloomington Stock Center, and the Developmental Hybridoma Studies Bank. I acknowledge the gracious support of Gerald Rubin and the Damon Runyon Cancer Research Foundation, DRG 1632.


  • Communicating editor: T. C. Kaufman

  • Received September 9, 2002.
  • Accepted January 8, 2003.


View Abstract