Zinc-finger paralogues tsh and tio are functionally equivalent during imaginal development in Drosophila and maintain their expression levels through auto- and cross-negative feedback loops


  • José Bessa,

    1. Centro Andaluz de Biología del Desarrollo (CABD), CSIC-Universidad Pablo de Olavide, Campus UPO, Seville, Spain
    Search for more papers by this author
  • Laura Carmona,

    1. Centro Andaluz de Biología del Desarrollo (CABD), CSIC-Universidad Pablo de Olavide, Campus UPO, Seville, Spain
    Search for more papers by this author
  • Fernando Casares

    Corresponding author
    1. Centro Andaluz de Biología del Desarrollo (CABD), CSIC-Universidad Pablo de Olavide, Campus UPO, Seville, Spain
    2. Instituto de Biologia Molecular e Celular, Universidade do Porto, Oporto, Portugal
    • Centro Andaluz de Biologia del Desarrollo (CABD), CSIC-Universidad Pablo de Olavide, Campus UPO, Ctra. Utrera km1. Seville 41013, Spain
    Search for more papers by this author


teashirt (tsh) and tiptop (tio) are two Drosophila gene paralogues encoding zinc-finger transcription factors. While tsh is an important developmental regulator, tio null individuals are viable and fertile. Here, we show that tio and tsh have coincident expression domains in the imaginal discs, the precursors of the adult body, and that both genes show similar functional properties when expressed ectopically. Furthermore, tio is able to rescue the development of tsh mutants, indicating that both genes are functionally equivalent during imaginal development. Of interest, the transcriptional regulation of tio and tsh is linked by a negative feedback loop. This mechanism might be required to maintain a tight control on the total levels of tio/tsh and could help explaining why Drosophila keeps an apparently dispensable gene. Developmental Dynamics 238:19–28, 2009. © 2008 Wiley-Liss, Inc.


The Drosophila gene teashirt (tsh; Fasano et al.,1991) is the founding member of a family of evolutionary conserved Zn-finger genes encoding developmental regulators. During embryogenesis, tsh has been shown to participate in the patterning of the epidermis (Fasano et al.,1991; Roder et al.,1992; Andrew et al.,1994; de Zulueta et al.,1994; Gallet et al.,1998,2000; Robertson et al.,2004) and the development of the salivary glands (Henderson et al.,1999) and the gut (Mathies et al.,1994; Waltzer et al.,2001; Saller et al.,2002), and tsh null mutants are embryonic lethal (Fasano et al.,1991). The Drosophila genome contains also a tsh paralogue, tiptop (tio; Laugier et al.,2005). Both genes share an acidic N-terminal domain plus three similarly spaced Zn-fingers. In addition, tio has an extra fourth Zn-finger domain (Laugier et al.,2005). In early embryos, tio and tsh show mostly nonoverlapping expression domains, although, as embryogenesis proceeds, the coexpression of both genes becomes more extensive. In the embryo, tsh and tio negatively regulate each other. Thus, in the absence of one of the paralogues, there is increased expression of the other, even in domains where the up-regulated gene is usually not expressed. In addition, in tsh mutant embryos, tio up-regulation compensates partially the loss of tsh, so that the epidermal phenotype of tsh mutants increases if tio function is also eliminated (Laugier et al.,2005). Therefore, during embryogenesis tsh and tio are functionally similar. However, tio null mutants give rise to viable and fertile adult flies, at least under laboratory conditions. This indicates that, despite its embryonic expression, tio is essentially dispensable for normal development.

tsh is also expressed in the imaginal discs (Fasano et al.,1991), the larval precursors of the external body structures of the adult fly (Cohen,1993). tsh is expressed in the periphery of the leg discs, where it has been shown to be required for the development of its adult derivatives, the proximal leg segments (Abu-Shaar et al.,1999; Erkner et al.,1999; Wu and Cohen,2000). During wing disc development, tsh is expressed initially throughout the disc in first larval stage (L1), and its repression at the disc's center during L2 allows the specification of the wing. Later on, during L3, tsh expression becomes restricted to the prospective hinge, the region that articulates the wing to the body wall (Azpiazu and Morata,2000; Casares and Mann,2000; Soanes et al.,2001; Wu and Cohen,2002; Zirin and Mann,2004). Indeed, mutation of a hinge-specific tsh enhancer causes the aeroplane phenotype, characterized by flies displaying held out wings due to abnormal hinge morphology (Soanes et al.,2001). tsh is also expressed in the progenitor as well as in the atonal-expressing retinal precursor domains of the developing eye (Bessa et al.,2002). Gain-of-function experiments show that tsh can either promote or repress eye development (Pan and Rubin,1998; Singh et al.,2002; Bessa and Casares,2005), likely depending on the cellular context in which tsh is expressed and the duration of this expression (Bessa and Casares,2005). Loss-of-function experiments also show that tsh is required for eye development, although it has been noted that elimination of tsh in cell clones not always results in eye defects (Pan and Rubin,1998; Singh et al.,2002; Bessa and Casares,2005).

Here, we examine the expression of tio in imaginal discs and show that its expression domain largely overlaps that of tsh in these tissues. Ectopic expression of tio and tsh result in similar phenotypes. Moreover, tio is capable of rescuing to adulthood an embryonic lethal tsh mutant combination if driven with a tsh pattern. Together, these results indicate that tio and tsh are also functionally equivalent during imaginal development. However, overexpression of either tsh or tio in their expression domain by means of a tsh-GAL4 driver leads to lethality, which indicates that the tight control of the levels of tsh/tio is critical for normal development. In fact, we show that tsh and tio negatively auto- and cross-regulate their expression levels. Experiments in which the function of tsh is knocked-down in a tio null genetic background indicate that, as in the embryo, tio contributes to the development of the organs in which tsh is required.


tio and tsh Have Coincident Domains of Expression in Imaginal Discs

The lack of obvious phenotypes of tio null adult flies (Laugier et al.,2005) suggested that either tio is not expressed in the imaginal primordia or that it is expressed but not strictly required for normal development. Therefore, we checked the expression pattern of tio in third larval stage (L3) discs and compared it to that of tsh. tio transcription, monitored by in situ hybridization, is detected in the anterior part of the eye disc and in the proximal domains of leg and wing discs (Fig. 1A–C). This expression is very similar to that of tsh (Fasano et al.,1991). When colocalization studies were performed using antibodies against Tsh and Tio, we observed that both proteins were co-expressed (Fig. 1D–F). The coincidence of tsh and tio expression domains in imaginal discs contrasts with their expression in the embryo, where both genes show, in addition to overlapping domains, gene specific ones (Laugier et al.,2005).

Figure 1.

tio and tsh show similar expression patterns in imaginal discs. Expression of tio and tsh in the eye (left), wing (middle) and leg (right column) L3 imaginal discs. A–C: In situ hybridization for tio transcript. D–F: Imaginal discs co-stained for Tio (green) and Tsh (red) protein products detected with specific antibodies.

tio Shows Similar Functional Properties to tsh an Can Substitute it During Imaginal Development

As tio and tsh have been shown to be functionally similar during embryogenesis (Laugier et al.,2005), we decided to investigate if this was also the case during imaginal disc development. We performed three experiments to test this point.

First, we ectopically expressed tio using the dpp-GAL4 driver line, knowing that a similar ectopic expression of tsh generates ectopic eyes in the ventral head (Pan and Rubin,1998). Indeed, dpp-GAL4; UAS-tio flies also develop ectopic eyes in the ventral head region (Fig. 2B). Furthermore, similar to tsh (Bessa and Casares,2005), expression of tio in cells clones in the peripodial epithelium (PE) of the eye disc induces the expression of eya, which indicates the transformation of the PE cells toward eye fate (Fig. 2A).

Figure 2.

Ectopic expression of tio in the eye-antennal disc results ectopic eye tissue. A: Confocal z-section through an eye disc containing tio-expressing clones (arrowheads) in the peripodial epithelium (PE), marked with anti-Tio antibody (red). The Tio-positive cells express eya (green), detected by an antibody. Anti-Hth staining (blue) marks the PE. B,C: Ectopic eye induction generated by ectopic expression of tio (B) and tsh (C) driven by the dpp-GAL4 line.

Second, we tested if tio shared another functional property with tsh: its ability to maintain homothorax (hth) expression in the undifferentiated precursors of the eye disc when ectopically expressed (Bessa et al.,2002). To determine whether tio was able to mimic this function we generated ectopic tio+ clones and found that Hth was cell-autonomously maintained in the precursor region of the eye disc, similar to when tsh is ectopically expressed (Fig. 3A). In these same experiments, we also found that tio's ability to maintain Hth expression is not restricted to the eye disc, but happens in regions of other discs where hth expression is down-regulated during development (Fig. 3B–D), such as in the distal regions of the antennal (Casares and Mann,1998) and wing discs (Casares and Mann,2000; Wu and Cohen,2002). Furthermore, nuclear Extradenticle (Exd) was always associated with this Hth maintenance (not shown). Interestingly, we found that, while ectopic expression of tio maintains Hth protein, hth transcription is not always so, as monitored by a hth-LacZ (Fig. 3), a bona-fide enhancer trap reporter of hth transcription (Rieckhof et al.,1997). Because there was no information about whether tsh maintained the expression of hth, we performed similar experiments, but now inducing ectopic tsh-expressing clones in a hth-lacZ background. In these experiments, ectopic tsh expression was able to maintain Hth protein outside its normal domain without inducing hth transcription (Fig. 4A,B). Despite the ability of tsh and tio to maintain Hth expression, clones expressing a ds-tsh (RNAi) construct, that reduce Tsh protein to undetectable levels, did not affect Hth protein accumulation in the normal domain of hth expression of the eye disc (Fig. 4C). One interpretation of this result is that tsh does not regulate Hth levels in the eye disc. Alternatively, it might be that tio suffices to perform tsh functions when the latter is knocked-down.

Figure 3.

tio-expressing clones maintain Hth protein, but not always hth-LacZ. A–D: Eye (A), antennal (B), wing (C), and leg (D) imaginal discs of hth-LacZ larvae containing clones are shown. Clones were induced during L1–L2 and marked by high levels of Tio protein expression. With the imaging conditions used, the endogenous Tio signal is barely detectable. Discs were also stained with anti-Hth (blue) and anti-beta galactosidase (green). Arrowheads point to Tio-clones that show strong Hth signal but low or absent expression of hth-LacZ.

Figure 4.

tsh gain- and loss-of-function clones. A,B: Clones of ectopic tsh expression in the antennal (A) and wing (B) discs in a hth-LacZ background. Clones were induced during L1–L2 and marked by the absence of green fluorescent protein (GFP, green). Discs are also stained for Hth and beta-galactosidase. Arrowheads point to Tsh-clones that show strong Hth signal but low or absent expression of hth-LacZ. C: Clone expressing a ds-tsh (RNAi) construct spanning the region of overlap between Hth (blue) and Tsh (green) in a hth-LacZ larvae. The clone is detected by the loss of Tsh staining (white outline). Neither the Hth protein levels nor the beta-galactosidase signal suffer detectable changes.

And third, if tio were functionally equivalent to tsh, one might expect tio to rescue a loss of tsh function if expressed in the tsh domain. To test this point, we generated embryos of the genotype tsh8/tsh-GAL4; UAS-tio. While the heteroallelic mutant background tsh8/tsh-GAL4 is embryonic lethal, tsh8/tsh-GAL4; UAS-tio individuals reach late pupal stages. The resulting pharates show an almost normal adult body pattern, with just mild defects in proximal segments of the legs, notum and abdomen (Fig. 5A,B). This result shows that indeed tio can, at least partially, replace tsh function during the formation of adult structures. We confirmed this point by testing if tsh and tio interacted genetically. For that, we generated two genotypes, one in which tsh function was knocked-down throughout the adult body by inducing multiple clones of ds-tsh(RNAi) (“tshKD”), and another in which, in addition, tio function was removed (“tio-tshKD”; see the Experimental Procedures section). tshKD females eclosed with a consistent held-out wing and bent-down haltere phenotype, resembling the tsh hypomorphic mutant aeroplane (Soanes et al.,2001), which indicates a high sensitivity of the proximal structures of wings and halteres to the loss of tsh. Males died within the pupal case, so the wing phenotype could not be assessed, but they showed no obvious additional phenotypes (n > 50 for each sex). In contrast, both female and male tio-tshKD individuals died before eclosion, sometimes before cuticle formation. More developed pupae showed several developmental defects that had been described for tsh nulls, such as fusion of proximal leg segments (Fig. 5C,D) (Erkner et al.,1999; Wu and Cohen,2000) that we never observed in tio-tshKD flies. These results indicate that during imaginal development tio can partially compensate for the loss of tsh.

Figure 5.

Expression of tio in the tsh domain rescues an embryonic lethal tsh mutant genotype up to late pupal stages. A,B: Dorsal (A) and ventral (B) views of a tsh8/tsh-GAL4; UAS-tio/+ pharate adult, dissected out of its pupal case. Body structures are almost normal, with the exception of proximal legs defects (not visible at this magnification) and incomplete fusion of dorsal abdominal terguites. C,D: Proximal segments of the first thoracic leg of tshKD (C) and tio-tshKD (D) males. In both cases, most or all the tissue shown is mutant, as marked by the light (yellow) bristles. Coxa (Co), trocanter (trc) and femur (Fe) segments are well formed in tshKD animals (C), while tio-tshKD flies (D) show fused coxa and trocanter segments with abnormal morphology (asterisk).

tio and tsh Are Engaged in Negative Auto- and Cross-regulatory Feedback Loops

Of interest, the genotypes tsh-GAL4 UAS-tio or tsh-GAL4 UAS-tsh were never viable, inducing lethality before pupal stages. This observation suggests that Tsh and/or Tio levels are critical for normal development. Therefore, it would be predicted the total concentration of Tio and Tsh present in a cell to be tightly regulated. Transcriptional cross regulation between tio and tsh could ensure that their protein levels are kept constant as has been shown in the embryo (Laugier et al.,2005). Here we tested whether similar negative cross regulation between these genes exists in the imaginal discs. To test this, we generated clones overexpressing tio and analyzed the effects on tsh transcription, using a tsh-Z reporter line, and on protein levels, using an anti-Tsh antibody. tio overexpression caused a dramatic down-regulation of tsh transcription and protein in the eye, wing and leg imaginal discs (Fig. 6A–C). Conversely, tsh overexpression along the AP border of the wing disc using a dpp-GAL4 driver line down-regulated Tio (Fig. 6D). On the other hand, loss of Tsh induced by ectopic expression of a tsh RNAi resulted in a cell-autonomous increase of Tio levels in discs (Fig. 7A,B). These results indicate that tio and tsh negatively regulate each other.

Figure 6.

Overexpression of tio or tsh represses each other. A–C: Overexpression clones of tio (red) repress tsh transcription (monitored by a tsh-Z reporter; white, and blue in the merged image) and Tsh protein production (green) in eye (A), wing (B), and leg (C) imaginal discs. In all cases the overproduction of tio represses tsh at the transcriptional and protein expression levels. D: High levels of tsh (green) expression, driven by a dpp-GAL4 driver, represses tio (red) in the notum region of the wing imaginal disc, a place where both genes are normally co-expressed (dashed lines mark the region of tsh overexpression).

Figure 7.

Down-regulation of tsh through an RNAi construct results in increased expression of tio. A,B: Flip-out clones expressing a ds-tsh (RNAi) construct in eye (A) and wing (B) imaginal discs. These clones are marked by the lack of CD2 (red) and show undetectable levels of Tsh antigen (blue). The levels of Tio antigen (green) are higher within the clones than in surrounding cells (clones outlined with dashed white lines).

Because tio and tsh regulate each other transcriptionally, and taking into account their molecular similarity, it is also likely that each of these genes auto-regulates its own expression levels. We tested this point for tsh, by generating tsh loss- and gain-of-function clones (Fig. 8) in a tsh-lacZ background. tsh-lacZ expression, which reports on the transcription of tsh, is up-regulated in tsh-knock down clones in eye, wing, and leg imaginal discs (Fig. 8C,D). Conversely, ectopic tsh expression causes the down-regulation of tsh-lacZ (Fig. 8A,B), confirming that the levels of tsh itself are controlled by a negative autoregulatory loop.

Figure 8.

tsh negatively regulates its own expression. A,B: Gain-of-function and (C,D) loss-of-function tsh flip-out clones induced in a tsh-Z background. tsh+ clones (A,B), marked by the absence of GFP (green), show decreased tsh-Z expression (red), detected by anti beta-galactosidase staining. Clones in the eye (A) and wing (B) discs are shown (arrowheads). ds-tsh (RNAi) clones (C,D), detected by absence of Tsh immunoreactivity (green), show an increase in tsh-Z expression (red), detected by anti beta-galactosidase staining (arrowheads). In all experiments, Hth staining was included to mark the cells' nuclei in the tsh domain.


Tio and Tsh share considerable homology on the sequence, organization and spacing of their acidic and Zn-finger domains. Here, we have shown that tsh and tio have coincident patterns of expression in imaginal discs, both genes exhibit similar biological functions when overexpressed, and also that tio can partially replace tsh function during adult development. Expression of tio within the tsh domain is able to rescue the embryonic lethality of a strong hypomorphic tsh mutant condition, allowing the survival of individuals up to late pupa stages. In these pharates, most of the external adult structures differentiate properly. This result indicates that tio is functional equivalent to tsh in most, if not all, of its roles. Molecularly, both Tsh and Tio maintain Hth protein in regions where its transcription is already turned off. It was previously shown that Tsh and Hth have the potential to interact directly (Bessa et al.,2002) and because of their molecular and functional similarity, this could also be the case for Tio. This interaction between Hth and Tio and/or Tsh could result in the stabilization of the Hth protein.

Together with previous reports on the embryonic function of tio and tsh (Laugier et al.,2005), our results show that both genes have, on the whole, a common function.

Therefore, the dispensability of tio function for the development of the fly (Laugier et al.,2005), at least under laboratory conditions, could be explained by the sufficiency of tsh to carry out all developmental functions that would normally be performed by the pair of paralogues. One important component to this sufficiency could reside in the ability of the tsh/tio system to regulate the total levels of their proteins. The negative cross-regulation we have described would result in the compensatory up-regulation of tsh expression if tio function is missing. Intuitively, it may seem that the tsh/tio double repression circuit (in which each gene negatively regulates its own, as well as the other's expression) should result in the extinction of the transcription of one of the two genes. Nevertheless, mathematical modeling indicates that double repression circuits of two components are indeed compatible with a single steady state in which the two genes are expressed at low levels (R. Guantes and J.F. Poyatos, personal communication).

The dispensability of tio raises the question of why Drosophila has two equivalent genes. tsh genes also exist as a family of paralogues in mammals (Manfroid et al.,2004). However, these duplications have arisen independently in the vertebrate and invertebrate lineages. In insects, only one tsh/tio gene is found in T. castaneum, A. mellifera, A. gambiae (Shippy et al.,2008) and Glossina morsitans morsitans (R. Bao and M. Friedrich, personal communication), but present in the genomes of all 12 Drosophila species sequenced. These data indicate that the duplication leading to the tsh/tio pair arose only after the separation of the Drosophila and Glossina lineages, between 60 and 140 million years ago. The fact that the embryonic pattern of expression of the single Tc-tiotsh gene in the flour beetle roughly resembles the summation of the patterns of tsh and tio in the Drosophila embryo points to a possible subfunctionalization of tsh and tio after the duplication (Lynch and Force,2000), as discussed by Shippy et al. (2008). This subfunctionalization would be quantitative during imaginal disc development, because although tsh and tio are expressed in coincident domains and their protein products are functionally equivalent, loss of tsh cannot be fully compensated by tio. An additional explanation for the retention of tio might lie in the tsh/tio regulatory relationship. Negative feedback loops, as the one engaging tsh and tio, are commonly associated to mechanisms of noise reduction, which are capable of attenuating fluctuations in gene products (Kaern et al.,2005). Fast developers, such as Drosophila, may be especially vulnerable to random or environmentally induced gene expression fluctuations. In fact, the Drosophila genome is particularly rich in gene duplications involving, in many instances, transcription factors (Richards et al.,2008). It would be thus interesting to investigate whether mechanisms of mutual negative cross-regulation are pervasive among Drosophila paralogues.


Genotypes and Genetic Manipulations

Larvae were raised at 25°C, unless otherwise indicated. tsh1 (tshZ; Fasano et al.,1991) and hth06762, (hth-Z; Casares and Mann,2000), were used as reporters for tsh and hth transcription. Other mutations were the tio null mutation tio473 (Dockendorff et al.,2002; Laugier et al.,2005) and the null mutation tsh8 (Fasano et al.,1991).

For targeted missexpression we used the UAS/GAL4 system (Brand and Perrimon,1993). Lines used were as follows: UAS-tsh (Gallet et al.,1998); UAS-tio (Laugier et al.,2005), dpp-GAL4 (Staehling-Hampton et al.,1994), tsh-GAL4 (Wu and Cohen,2000), UAS-ds-tsh (RNAi) (Bessa and Casares,2005).

Ectopic expression clones were generated randomly in imaginal discs by heat shocking L1–L2 larvae (24–72 hr after egg laying [AEL]) for 30 min at 35.5°C of the genotypes: yw hsFlp122; tub> GFP, y+> GAL4 (Zecca and Struhl,2002); UAS-tsh/ SM6TM6B for tsh-expressing clones; and y hsFlp122, act> hsCD2> GAL4 (Struhl and Basler,1993); UAS-tio for tio-expressing clones. In some experiments, gene expression reporters were introduced in the genotype by standard genetic methods. Clones were marked negatively by the absence of GFP or CD2, or positively by detection of Tsh or Tio antigens. For CD2 induction, larvae were subject to a 30-min 37°C heat-shock followed by a 30-min recovery period at room temperature just before dissection.

tsh-knock-down clones were induced between 28 and 72 hr AEL in larvae of the genotype yw hs-Flp122; act> y+> GAL4, UAS-lacZ; UAS-tshRNAi (Bessa and Casares,2005).

The “tsh knock down” (tshKD) and “tio-; tshKD” conditions were generated in larvae of the genotypes yw hsFlp12; tio473/+; UAS-ds-tsh/act>y+>GAL4; and yw hsFlp12; tio473/tio473; UAS-ds-tsh/ act>y+>GAL4, respectively, by heat-shocking larvae for 1 hr at 37°C twice, one at 24 and another at 48 hAEL. Mutant tissue can be visualized in the adults/pharates as yellow (y) tissue. With this regimen, the extent of y-marked tissue ranges approximately between 70 to 90%, indicating a broad expression of the ds-tsh construct. Adult flies or pharates dissected out of the pupal case were stored in 70% ethanol. This preservation method bleaches the eye pigmentation. Specimens were directly photographed under a binocular scope or dissected and mounted in Hoyer's solution: Lactic Acid (1:1) for microscopic examination.


Antibodies used are as follows: mouse anti–β-galactosidase (Sigma), mouse anti-CD2 (Serotec), guinea pig anti-Hth (Casares and Mann,1998), rabbit anti-Tsh (Wu and Cohen,2000), rat anti-Tio antibody (used 1/100; Laugier et al.,2005), and rabbit anti-Exd (Gonzalez-Crespo and Morata,1995). The monoclonal anti-Eya antibody (Bonini et al.,1993) was from the DSHB, University of Iowa. Anti-mouse, -rabbit, and -guinea pig secondary antibodies, conjugated with Alexa 488, 568, or 647 are from Molecular Probes. Green fluorescent protein (GFP) signal was directly detected. Images were obtained with a SP2-AOBS Leica confocal system and processed with Adobe-Photoshop.

In Situ Hybridization

The tiptop EST clone GH27226 (BDGP) was used to synthesize a tio DIG labeled RNA probe. The DNA was linearized with EcoRI and the RNA probe was synthesized with an Sp6 RNA-polymerase using a digoxigenin (DIG) RNA labeling kit (Roche #1175025). The in situ protocol used was that of (Tautz and Pfeifle,1989) with slight modification, and was carried out after standard imaginal discs fixation.


We thank L. Fasano and S. Kerridge for the anti-Tio antiserum and fly strains, J.F. Poyatos and M. Friedrich for communicating results before publication, M. Friedrich for comments on the manuscript, and the CABD advanced light microscopy facility for technical support. J.B. is a postdoctoral fellow of the Portuguese Fundação para a Ciência e a Tecnologia. CABD is institutionally supported by CSIC, Universidad Pablo de Olavide and Junta de Andalucía.