Eye suppression, a novel function of teashirt, requires Wingless signaling

The compound eye of adult Drosophila, which consists of a hexagonal array of about 800 ommatidia, develops from the larval eye disc. The eye disc differentiates progressively in a posterior to anterior direction, with a morphogenetic furrow (MF) marking the front of the differentiation wave. A small number of genes (eyeless, sine oculis, eyes absent and dachshund) encoding nuclear factors have been identified to be important for eye formation. Loss-of-function mutations in these genes block eye development, while targeted expression, alone or in combination, can induce ectopic eyes (for reviews, see Desplan, 1997; Treisman, 1999; Heberlein and Treisman, 2000). However, several genes are known to block eye formation. homothorax (hth), which encodes a homeodomain protein (Rieckhof et al., 1997; Pai et al., 1998; Kurant et al., 1998), is expressed in the anterior margin of the eye disc. Mutant hth clones cause ectopic eye formation in the ventral head, whereas ectopic hth expression in the eye field blocks MF initiation and progression (Pai et al., 1998; Pichaud and Casares, 2000). Signaling by Wingless (WG), expressed along the anterolateral margins, also blocks MF initiation and progression (Ma and Moses, 1995; Treisman and Rubin, 1995). extra macrochaetae (emc) and hairy (h), both of which encode transcription factors, are expressed anterior to the MF and act redundantly to block MF progression (Brown et al., 1995). teashirt (tsh) encodes a nuclear protein with zinc-finger motifs (Fasano et al., 1991). It is involved in embryonic trunk segmental identity (Fasano et al., 1991; Roder et al., 1992; de Zulueta et al., 1994; Alexandre et al., 1996) and midgut morphogenesis (Mathies et al., 1994), and confers proximal identity in leg development (Erkner et al., 1999; Wu and Cohen, 2000). A role for tsh in eye development was initially suggested because the eyes of flies trans-heterozygous for tsh and gain-of-function Antennapedia mutations were reduced and partially transformed to head cuticle (Bhojwani et al., 1997). Pan and Rubin (Pan and Rubin, 1998) showed that targeted misexpression of tsh could induce eyeless (ey) expression and generate ectopic eyes in the antenna. In this study, we report that tsh has a novel function in suppressing eye development. This eye suppression function is achieved through the induction of hth and requires WG signaling.

Although the cellular composition of each ommatidium is identical, their spatial arrangements show mirror symmetry over the dorsoventral (DV) midline (equator) in the eye (Wolff and Ready, 1993). Early eye primordia is subdivided into dorsal and ventral compartments (Cho and Choi, 1998; Dominguez and de Celis, 1998; Papayannopoulos et al., 1998; Cavodeassi et al., 1999). Many genes exhibit DV asymmetry in their expression and/or function in the eye. Some genes (e.g. wg) have symmetrical DV expression, but are regulated differently or have DV differential functions. For example, the dorsal-specific expression of the iro-C genes (McNeill et al., 1997; Dominguez and de Celis, 1998; Cavodeassi et al., 1999; Cavodeassi et al., 2000) requires WG and Hedgehog signaling (Heberlein et al., 1998; Cavodeassi et al., 1999). The dorsal, but not ventral, wg expression in turn requires the GATA factor pannier (Maurel-Zaffran and Treisman, 2000; Lee and Treisman, 2001). WG induces mirr in dorsal and contributes to dorsal eye fate (Heberlein et al., 1998), whereas wg on the ventral margin can induce and maintain hth, a negative regulator of eye (Pai et al., 1998), and suppress ventral eye fate (Pichaud and Casares, 2000). We found that although tsh was expressed in a DV symmetrical pattern in the eye disc, its function in the eye showed DV asymmetry: tsh suppresses eye development in the ventral region, but promotes eye development in the dorsal region. The effect of tsh is probably on both early eye disc growth and photoreceptor differentiation. We also show that misexpression of tsh has DV differential effects in the antennal disc, but not in the wing disc. These disc- and position-dependent effects are presumably due to the involvement of additional factors.

We used the GAL4/UAS system for the targeted misexpression (Brand and Perrimon, 1993). tsh-GAL4 (Shiga et al., 1996), dpp-GAL4 (Staehling-Hampton and Hoffmann, 1994), ey-GAL4 (Hazelett et al., 1998), bi-GAL4 (Calleja et al., 1996), UAS-tsh (Gallet et al., 1998), UAS-hth (Pai et al., 1998), UAS-fluΔarm (Zecca et al., 1996), UAS-wg (Azpiazu and Morata, 1998), UAS-sgg (Hazelett et al., 1998) and UAS-dTCF^ΔN (van de Wetering et al., 1997) were used. The flies were cultured at three different temperatures: 18°C, 25°C and 29°C to sample the effect of different induction level.

w; P(Act>y⁺>GAL4)25 P(UAS-GFP^S65T)/CyO (Ito et al., 1997) and y w hsFLP¹²² (Struhl and Basler, 1993) were used for generating expression clones. All other stocks were constructed using these stocks by suitable genetic crosses. Embryos were collected at 12 hours interval at 25°C, and subjected to a single 1 hour heat shock at 37°C at about 24 hours after egg laying (AEL) or as indicated. The larvae were transferred to 25°C for recovery and further development.

We used the temperature-sensitive allele wg^IL114 (Nusslein-Volhard et al., 1984; Treisman and Rubin, 1995). The F₁ progeny of the genotypes w; wg^IL114; dpp-GAL4 /SM6-TM6B and w; wg^IL114; UAS-tsh were grown at 17°C and shifted to 29°C at various developmental stages for a period of 24 hours and returned to 17°C for further development. The eye phenotypes were studied in imaginal discs dissected from Tb⁺ third instar larvae or in pharate adults.

tsh is located on the second chromosome at 40A, too close to the FRT(40A) for recombination onto the FRT chromosome to generate loss-of-function clones (Xu and Rubin, 1993). A null allele of tsh, tsh⁸ (Fasano et al., 1991) was used to generate loss-of-function clones by X-ray irradiation following the protocol of Wu and Cohen (Wu and Cohen, 2000). A viable enhancer trap insertion, tshA8, with the P[lacW] inserted near the tsh locus (Sun et al., 1995), served as a marker for the tsh⁺ chromosome. tsh⁸ clones were generated by irradiating tsh⁸/tshA8 larvae (Fig. 4A), from 6 hours egg collections, with 4000 rads of X-ray at 48-72 hours or 72-90 hours AEL. In adult eyes, tsh⁸ clones were detected by the loss of eye color, which is dependent on the mini-white reporter gene in tshA8. As mini-white in tshA8 causes eye color only in the anterior half of the eye, corresponding to the tsh expression domain, clones in the tsh non-expressing posterior half of the eye cannot be detected. Loss-of-function clones in the posterior region of the third instar eye disc cannot be marked, but only deduced from their effects.

Eye-antennal and wing imaginal discs were dissected from wandering third instar larvae in 1×phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde (in PBS) for 20 minute. They were washed three times with PBST (1×PBS+2% Triton-X-100) for 10 minute each and blocked in 10% normal goat serum for 1 hour. The discs were incubated overnight at 4°C in rat anti-ELAV (1:200) (Developmental Studies Hybridoma Bank) and one of the following primary antibodies: rabbit anti-β-GAL (1:800) (Cappel), mouse anti-WG (1:20) (Steve Cohen), rabbit anti-HTH (1:200) (Pai et al., 1998) or rabbit anti-EY (1:200) (Uwe Walldorf). The discs were washed in PBST twice for 10 minute each and blocked again for 30 minutes in 10% goat serum. Secondary antibodies (Jackson Laboratories) were donkey anti-rat IgG conjugated to Cy5 (1:200), donkey anti-rabbit IgG conjugated to Cy3 (1:400) or goat anti-mouse IgG conjugated to FITC or Alexa Fluor 488 (1:200). The discs were incubated with secondary antibodies for about 2 hour and washed in PBST for 10 minute. The discs were mounted in DABCO (Sigma) mountant in 90% glycerol and photo-documented on a Zeiss LSM510 confocal microscope.

Ectopic induction of tsh under dpp-GAL4 (dpp>tsh) could occasionally cause the formation of an ectopic eye at the base of the antenna (Fig. 1a, arrowhead), as previously reported (Pan and Rubin, 1998). In addition, about 8% (17/212) of the dpp>tsh flies showed a split-eye phenotype. Nearly 42% (89/212) of the dpp>tsh individuals arrested and died at the white pupal stage. Their body size was about 50% larger than normal. These larvae have extended larval period (by 2-3 days) and their eye-antennal discs showed overgrowth and distorted morphology. These discs showed both variable ectopic HTH induction and eye suppression phenotypes, from a small group of ELAV-positive photoreceptor cells (Fig. 1b) to near complete absence of ELAV-positive cells (Fig. 1c). Occasionally, HTH induction extended from the posterior margin towards the MF and the endogenous eye field was split into two (Fig. 1b). The frequency of splitting of the eye field was comparable with the split-eye phenotype in the adult flies. These observations suggested that the dpp>tsh split-eye phenotype was due to suppression of the eye fate by tsh, resulting in the splitting of the endogenous eye field. Consistent with this interpretation, when two copies of UAS-tsh were driven by the dpp-GAL4 (dpp>2Xtsh), the ventral half of the endogenous eye was completely absent in all flies (Fig. 1d), indicating that higher levels of tsh expression is capable of completely suppressing ventral eye development. Ectopic induction of tsh by ey-GAL4 (ey>tsh), which is expressed in the embryonic eye primordium and in the eye disc (Hazelett et al., 1998), caused a complete suppression of eye development (Fig. 1e,f) with 97% penetrance. Only 2.4% (2/83) of the ey>tsh flies have very small eyes.

In dpp>tsh, the frequency of HTH induction correlated with the frequency of eye suppression (see above), suggesting that hth may be responsible for the eye suppression. dpp>tsh in a heterozygous hth mutant background resulted in the reduction of the split-eye frequency to 0.6% (2/164; from 8% in dpp>tsh) in pharate adults (Fig. 1g). The pupal lethality was also reduced to 16% (27/164) from 42%. These results suggested that these phenotypes are hth-dependent. dpp>hth completely suppressed eye development (Pai et al., 1998). Co-expression of tsh and hth (dpp>tsh+hth) also completely suppressed eye development and induced no antennal eye (Fig. 1h), suggesting that hth acts downstream of, or in parallel to, tsh.

As ectopic tsh could regulate hth expression, we compared their endogenous expression in imaginal discs. Expression of tsh was examined using tsh-GAL4 (Shiga et al., 1996)-driven UAS-GFP (tsh>GFP). In the eye disc, tsh expression could be detected as early as first larval instar in the entire disc proper, overlapping with hth and the pro-eye gene eyeless (ey) (Fig. 2a). In the late second instar eye disc (Fig. 2b), tsh>GFP expression retracts anteriorly and occupied nearly three quarters of the disc. hth expression also retracts anteriorly, as also reported by Pichaud and Casares (Pichaud and Casares, 2000). EY is also expressed in the same region (Fig. 2b) (Halder et al., 1998). In early third instar eye discs (Fig. 2c), tsh expression regresses to the anterior two-thirds of the disc. hth expression is restricted to the anterior margin in a 10- to 15-cell wide domain. tsh and hth expression overlaps in a 3- to 4-cell wide stripe. EY expression (Fig. 2c) (Halder et al., 1998) largely overlapped with tsh. In late third instar eye disc (Fig. 2d), tsh>GFP expression was anterior to the MF and was similar to the expression pattern determined by tsh-lacZ, anti-TSH antibody and in situ hybridization (Sun et al., 1995; Bhojwani et al., 1997; Pan and Rubin, 1998). The co-expression of tsh and hth during the early phase of eye disc development is consistent with the finding that tsh induces hth expression.

In late third instar wing disc, tsh is expressed in a proximal ring around the wing pouch and in most of the notum (Fig. 2e) (Sun et al., 1995; Bhojwani et al., 1995; Casares and Mann, 2000; Azpiazu and Morata, 2000), largely overlapping with, but broader than, hth expression (Fig. 2e) (Pai et al., 1998; Azpiazu and Morata, 2000; Casares and Mann, 2000). EY is not expressed in the wing disc (Halder et al., 1995). In the antennal disc, tsh is expressed weakly in an anteroproximal region (Bhojwani et al., 1997; Pan and Rubin, 1998), while hth is expressed in the proximal region (Rieckhof et al., 1997; Pai et al., 1998; Casares and Mann, 1998).

dpp>2Xtsh showed suppression only of the ventral eye (Fig. 1d). We tested this ventral bias using bi-GAL4. bi>GFP is expressed at the dorsal and ventral margins of the eye disc (Fig. 3a). bi>tsh resulted in HTH induction and ELAV suppression only in the ventral region of the eye disc (Fig. 3b, arrow), and the absence of ventral eye in pharate adults (Fig. 3c, arrow). By contrast, bi>hth resulted in eye suppression in both dorsal and ventral regions in the eye disc (not shown) and in the adult (Fig. 3d, arrows).

Clonal induction of tsh expression [abbreviated Act>tsh, as the expression is driven by an Actin promoter using the flp-out system of Ito et al. (Ito et al., 1997)] could induce HTH and suppressed photoreceptor development in the eye disc, but only along the ventral margin (Fig. 3e, arrowhead). The induction of hth is at the transcription level, because an enhancer trap (hth^1422-4) lacZ reporter is also induced (not shown). Act>tsh in the dorsal margin of the eye disc, unlike the ventral clones, could cause overgrowth of the eye cells (Fig. 3e, arrow). Although the Act>tsh clones were not marked in the adult eye, adult flies with clonal induction of tsh showed ventral eye suppression (Fig. 3f, arrow) and dorsal eye enlargement (Fig. 3g, arrow). Internal Act>tsh clones located away from the margin in both the dorsal and ventral eye (not shown), irrespective of their size, did not affect the eye fate.

The null alleles of tsh are embryonic lethal and the available hypomorphic alleles do not show any eye defects (Bhojwani et al., 1997; Pan and Rubin, 1998). We generated loss-of-function clones of tsh⁸, a null allele, by X-ray irradiation (Fig. 4a) (Wu and Cohen, 2000) at different time windows beginning from 48 to 96 hours after egg laying (AEL). Eye phenotypes were observed only in flies irradiated around 52-64 hours AEL. tsh⁸ clones located in ventral margin of the adult eye caused ventral eye enlargements (Fig. 4c,d, arrow). Internal ventral clones did not significantly affect eye development (not shown) (Pan and Rubin, 1998). tsh⁸ clones in the posterior region of the third instar eye disc could not be marked, because at this stage tsh (and the tshA8 reporter) expression has already retracted from this region of early expression. However, after clone induction, ventral enlargements of the eye field, corresponding to the adult eye phenotype, were seen in the eye discs (Fig. 4e,f, arrows). In rare cases ectopic ventral eyes were also observed (Fig. 4g, arrow). tsh⁸ clones located in the dorsal eye suppressed eye fate and caused eye-to-cuticle transformation (Fig. 4h,i, arrow). One eye disc had a complete absence of the dorsal eye field (Fig. 4j, arrow). These phenotypes were not found when wild-type larvae were similarly treated by X-ray irradiation. These results suggested that the normal function of tsh is to suppress eye fate in the ventral eye and to promote eye fate in the dorsal eye during early second instar.

Pichaud and Casares (Pichaud and Casares, 2000) reported that hth and wg are involved in a positive feedback loop only in the ventral eye disc, but not in the dorsal region. Gallet et al. (Gallet et al., 1998; Gallet et al., 1999) showed that TSH binds ARM, a downstream component of WG signaling. We therefore, checked the role of WG signaling in the tsh-mediated HTH induction.

Clonal induction of tsh together with a constitutively activated ARM (Zecca et al., 1996) caused ectopic induction of HTH and suppressed eye development both in dorsal (Fig. 5a, arrow) and ventral domains, and both in marginal and internal regions of eye disc (Fig. 5a). Some of these Act>tsh+arm clones were also associated with tissue overgrowth as seen in adult eye (not shown). These phenotypes were similar to ectopic hth expression with the exception of tissue overgrowth (Azpiazu and Morata, 2000; Casares and Mann, 2000; Goto and Hayashi, 1999; Jaw et al., 2000; Pai et al., 1998). Act>arm showed variable phenotypes: an internal ventral clone could suppress eye (Fig. 5b, arrow), whereas a clone at the posterior margin could not suppress eye fate (Fig. 5b, arrowhead). dpp>tsh+arm resulted in complete suppression of eye (not shown), similar to the dpp>hth phenotype (Pai et al., 1998). In bi>tsh+arm, eye is reduced in both dorsal and ventral margins in discs and in adults (not shown). Act>tsh+wg also resulted in HTH induction and eye suppression in dorsal eye disc (Fig. 5c). In bi> tsh+wg, the eye field was extremely reduced at both dorsal and ventral margins in discs (Fig. 5d) and in adults (not shown). These results suggested that WG signaling can collaborate with TSH for HTH-mediated suppression of eye fate.

The requirement of WG signaling in the tsh-mediated eye suppression was examined by co-expressing tsh with antagonists of WG signaling. dTCF^ΔN, a dominant negative form of dTCF, can block the WG signaling (van de Wetering et al., 1997). Shaggy zeste white-3 (SGG) also acts as an antagonist of WG signaling (Hazelett et al., 1998; Heslip et al., 1997). Act>dTCF^ΔN+tsh, unlike Act>tsh, failed to induce HTH and suppress eye development, irrespective of the dorsal or ventral domain, in both discs (Fig. 5e) and adults (not shown). Act>dTCF^ΔN did not induce HTH or suppress eye fate (not shown). As expected, bi>tsh+dTCF^ΔN did not show suppression of the eye fate both on the dorsal and ventral margins in the eye disc (not shown) and in adult (Fig. 5f). In ey>tsh+dTCF^ΔN and ey>tsh+sgg eye discs and flies, there was no eye suppression (not shown). Similarly, Act>tsh+sgg (Fig. 5g) and bi>tsh+sgg (not shown) did not induce HTH or suppress eye development both in the ventral or dorsal margin in the disc and in flies. Act>sgg did not suppress eye fate (not shown). These observations suggest that WG signaling is required for the ventral eye suppression mediated by tsh.

The temporal requirement of WG signaling was examined by misexpressing tsh in wg^ts mutant flies (see Materials and Methods). When the flies were shifted to the restrictive temperature 48-72 hours AEL, the frequency of split-eye phenotype caused by dpp>tsh was reduced to 2% (4/228). The frequency of white pupal lethality was reduced to 19% (43/228). Temperature shifts in other time windows did not affect the frequency and severity of the split-eye phenotype (not shown). This critical period corresponds to the second instar larval stage and is consistent with the above finding that the tsh function is required during this period.

To examine whether tsh has a direct effect on eye disc growth, we measured the relative frequency and size of Act>GFP and Act>tsh+GFP clones induced at the same time (early first instar and second instar). Act>GFP clones were equally abundant both in the dorsal and ventral eye (Fig. 6a), but Act>tsh+GFP clones were very scarce. The frequency of the dorsal Act>tsh+GFP clones (16 clones) was higher than that of the ventral Act>tsh+GFP clones (four clones). Twelve of the 16 dorsal Act>tsh+GFP clones showed overgrowth (Fig. 6b). By comparison, only one of the four ventral internal Act>tsh+GFP clones showed weak overgrowth (Fig. 6c). The other three ventral clones were smaller than the average Act>GFP clones (not shown). These results suggested that tsh is involved in growth regulation and has opposite effect in the dorsal and ventral region.

tsh also showed DV differential activities in the leg disc (Erkner et al., 1999). We examined whether similar DV differential activities of tsh, and the tsh-hth relationship, also occurs in the wing and antennal discs. Act>tsh in the wing disc induced HTH (Fig. 7a), as previously reported (Azpiazu and Morata, 2000; Casares and Mann, 2000). Induction of hth is at the transcriptional level (data not shown) (Casares and Mann, 2000). Unlike the eye and leg discs, there is no DV differential activity in the wing discs. HTH suppresses wg expression in the presumptive wing margin, while enhancing wg expression in the hinge region (Azpiazu and Morata, 2000; Casares and Mann, 2000). However, clonal induction of tsh, while inducing HTH, has no effect on WG in the wing pouch (Fig. 7b, arrow). These results suggested that TSH, in addition to inducing HTH in the wing, has another function: it prevents wg from being suppressed by HTH. dpp>tsh induced HTH along the AP compartmental boundary in the wing pouch and resulted in splitting of the wing pouch (Fig. 7c, arrow), as evident from splitting of the DV border-specific WG stripe. bi>tsh could induce HTH in the wing pouch (which spans the AP border) and splits the wing pouch (Fig. 7d, arrow), whereas bi>hth on its own could not split the wing field (Fig. 7e). These results again suggest that TSH has functions in addition to that of inducing HTH. The induction of HTH in bi>tsh wing disc showed no DV difference (Fig. 7d).

In the antennal disc, clonal induction of tsh in the ventral domain (Fig. 7f, arrow) caused a duplication of the antennal field as shown by duplication of the ventral WG expression domain (Fig. 7f). This phenotype could also be seen in an adult where the antennal segments distal to AN2 were duplicated (Fig. 7g, arrow). HTH is repressed within the Act>tsh clones (Fig. 7f). In the dorsal domain, the effect depends on the spatial location. bi-GAL4 reflects the expression pattern of optomotor-blind (omb), which is expressed in a dorsal sector (spanning the AP compartmental border), in the antennal disc (Fig. 3a). bi-tsh caused no obvious antennal phenotype (not shown), suggesting that tsh has no effect in this dorsal domain of the antennal disc. Act>tsh clone in the dorsoproximal region (Fig.7h, arrow) and near the border between the eye and the antenna discs (Fig. 7h, arrowhead) did not affect HTH level but could cause overgrowth in the posterior dorsal region (Fig. 7i). Induction of tsh by the dpp-GAL4, which drives expression in a dorsal sector at the AP border in the antennal disc could induce eye formation in the anteroproximodorsal region of the antenna (Fig. 1a) (Pan and Rubin, 1998). A tsh-expressing clone in the same anteroproximodorsal region also caused ectopic eye formation (Fig. 7j, arrow). dpp>tsh did not cause antennal duplication, unlike dpp>hth (Yao et al., 1999), consistent with the non-induction of HTH by tsh (Fig. 7h,i). These results indicate that TSH has differential functions along the DV axis in the antennal disc, similar to the eye disc.

We showed that ectopic expression of tsh could suppress photoreceptor development, while loss-of-function tsh clones induced ectopic eye formation. This novel function of tsh occurs only at the ventral margin of the eye disc. Pan and Rubin (Pan and Rubin, 1998) noted that targeted expression of tsh could induce ectopic eye formation in the antennal disc, but did not observe the eye-suppression phenotype. The discrepancy may be in that they induced tsh expression by insertional activation using a P element carrying a dpp disc-enhancer coupled with a hsp70 basal promoter (Pan and Rubin, 1998), while we drove UAS-tsh expression using a dpp-GAL4.

Interestingly, although tsh is expressed symmetrically in the dorsal and ventral halves of the eye disc, overexpressing tsh in these regions suppressed eye development in the ventral region, while promoted eye development in the dorsal region. Why would overexpressing tsh in a region where it is normally expressed caused phenotype reciprocal to the loss-of-function tsh mutant phenotype? It is possibly a dose effect, as the ectopic expression of two copies of tsh transgene caused stronger effect (Fig. 1d). The normal level of TSH may be balanced with some opposing forces for proper development, thus too little and too much of TSH will cause reciprocal effects. A similar case is WG, which is normally expressed in both dorsal and ventral margins. Reducing WG level caused ectopic MF formation (Ma and Moses, 1995; Treisman and Rubin, 1995), while raising WG level blocks MF initiation (Treisman and Rubin, 1995).

The eye-suppression function of tsh is accompanied by the induction of hth at the transcriptional level. Eye suppression is reduced when the hth dose is reduced, suggesting that HTH is the major mediator of tsh-induced eye suppression. This is consistent with the known role of hth as a repressor of eye development (Pai et al., 1998; Jaw et al., 2000; Pichaud and Casares, 2000). In the wing disc, tsh also induces HTH, but our results show that tsh has additional effects (e.g. protecting wg from suppression by HTH and splitting the wing pouch). Whether tsh has additional effects in the eye disc awaits further study.

The eye-suppression function of tsh requires WG signaling, as blocking WG signaling by co-expressing dTCF^ΔN or sgg with tsh, or overexpressing tsh in a wg^ts mutant at the non-permissive temperature blocked the suppression effect. The critical time for wg involvement is 48-72 hours AEL, corresponding to the second instar larval stage. At this stage, the expression patterns of tsh, hth and wg in the eye disc overlap considerably (Fig. 2b) (Pichaud and Casares, 2000; Royet and Finkelstein, 1997), consistent with their functional interaction.

TSH could induce HTH and suppress eye development only in the ventral margin of the eye disc. Internal Act>tsh clones had no eye-suppression effects. The restriction of eye suppression to the eye disc margin, where wg is expressed, suggests that tsh does not induce wg but requires high level WG signaling. Indeed, clonal expression of tsh internal in the eye disc does not induce WG expression (not shown). When TSH is co-expressed with WG or an activated ARM, eye suppression could occur away from the margin, possibly because higher level of WG signaling is provided by the ectopic expression. TSH also requires high level of WG to repress Ubx transcription in the embryonic midgut (Waltzer et al., 2001).

Ectopic expression of WG in the region ahead of MF induces HTH, while blocking WG signaling (by clonal expression of dTCF^ΔN) reduced HTH in the presumptive head region of the eye disc (Pichaud and Casares, 2000). These locations correspond to tsh expression domain, consistent with the TSH-WG collaboration. Act>hth clones could block MF initiation without inducing ectopic wg expression (Pichaud and Casares, 2000), also suggesting that hth acts downstream of WG. Thus, these results suggest that TSH collaborates with WG signaling to induce HTH to suppress eye development.

TSH and WG signaling also collaborate during embryonic development. TSH acts in the late phase of WG signaling to promote the naked cuticle cell fate of larvae (Gallet et al., 1998). TSH phosphorylation and nuclear accumulation is partially promoted by WG signaling (Gallet et al., 1998; Gallet et al., 1999). Hypophosphorylated TSH can bind directly to the intracellular ARM (Gallet et al., 1999). The effect of TSH overexpression on embryo development is dependent on the interaction with ARM (Gallet et al., 1999). TSH can also associate with SGG, an inhibitory component of WG signaling that promotes ARM degradation and acts downstream of SGG (Gallet et al., 1999). Whether the same molecular interaction operates in the eye disc awaits further study.

Based on the loss-of-function phenotype and overexpression phenotype, tsh suppresses eye development only in the ventral eye, while promoting eye development in the dorsal eye. The DV difference in TSH function is not likely to be due to wg, as wg is expressed in both dorsal and ventral margins, with even higher levels in dorsal parts (Ma and Moses, 1995; Treisman and Rubin, 1995). In a wg temperature-sensitive mutant, an ectopic MF initiates more on the dorsal side (Ma and Moses, 1995). WG signaling upregulates hth in both dorsal and ventral regions of the eye disc (Pichaud and Casares, 2000). Thus, wg can induce hth and suppress eye development in both ventral and dorsal margins, but through different mechanisms. TSH collaborates with WG signaling for eye suppression only in the ventral margin, but not in the dorsal margin. Whether WG requires other co-factors in the dorsal margin is not known.

tsh promoted eye development in the dorsal margin (Fig. 3b,e,g). When TSH is co-expressed with WG or an activated ARM, the dorsal enlargement is blocked (Fig. 5a,c,d). When WG signaling is blocked in bi>tsh+dTCF^ΔN (Fig. 5f) and bi>tsh+sgg (not shown), eye enlargement occurred in both dorsal and ventral sides. These results suggested that in the dorsal eye, WG signaling blocks eye development at a step downstream of tsh function.

Some dorsal- or ventral-specific factor(s) may determine the outcome of TSH function. One possible mechanism is by affecting the collaboration between TSH and WG signaling (ARM or SGG). Our preliminary results indicated that the dorsal-expressing auracuan (ara) and Delta (Dll) can confer the dorsal specificity, and the ventral-expressing Serrate (Ser) can confer the ventral specificity to TSH function (A. S. and Y. H. Sun, unpublished).

The DV differential effect of tsh also occurs in the leg and antennal discs, but not in the wing disc. In the leg discs, when away from the border between the proximal tsh-expressing and distal Dll-expressing cells, clonal tsh induction caused no effect in the dorsal domain, but affected cell adhesion property and patterning when in the ventral domain. In the distal region of the leg disc, where tsh is not expressed, clonal tsh induction can lead to TSH protein accumulation only in the ventral domain, because of WG signaling (Erkner et al., 1999). In the antennal disc, Act>tsh clones in the ventral domain could cause HTH repression and antenna duplication (Fig. 7g). In the dorsal domain, the effect of tsh misexpression depends on the location. In the omb expression region, which spans the dorsal AP border, there is no effect on HTH and on antenna development. But further away from the AP border, tsh dorsal misexpression could cause overgrowth (in the posterior-proximal region) and ectopic eye formation (in the anterior-proximal region). These disc- and position-dependent differences in tsh function suggest the involvement of additional factors in determining the functional outcome of TSH.

The critical period for eye suppression by tsh is in the second instar larval stage, based on tsh mutant clones and on misexpression of tsh in wg^ts background. At this time, morphogenetic furrow has not initiated and photoreceptor differentiation has not begun. tsh mutant clones induced in second instar caused enlargement in the ventral eye field and reduction of eye cells in the dorsal eye field. In the ventral overgrowth, not all cells have differentiated into photoreceptors. These results suggest that the primary effect of tsh function is on growth in the early eye disc. When the relative frequency and size of Act>GFP and Act>tsh+GFP clones were compared (Fig. 6), the results showed that tsh promoted growth in the dorsal and suppressed growth in the ventral region. A dorsal clone anterior to the MF showed overgrowth (Fig. 6b), suggesting that the effect can be a general growth promotion and not limited to differentiating retinal cells.

However, tsh⁸ mutant clones in the dorsal eye caused a transformation of eye cells into cuticle fate, suggesting that tsh also plays a role in promoting eye fate (in dorsal). This role is consistent with the finding that tsh could induce ectopic eye formation in antenna (Pan and Rubin, 1998). In the ventral eye disc, a role in directly suppressing photoreceptor fate is also supported by the finding of an isolated ventral eye field in the eye disc with tsh⁸ clone induction (Fig. 4g). This direct role is consistent with the ventral activation of hth, which can directly suppress photoreceptor differentiation (Pai et al., 1998). Thus, tsh can affect both the growth of the eye disc and the differentiation of photoreceptors.

We thank Shigeo Hayashi, Stephen Kerridge, Steve Cohen, Jessica Treisman, Mark Muskavitch, Sonsoles Campuzano, Kwang-Wook Choi and the Bloomington Stock Center for Drosophila strains; Steve Kerridge, Steve Cohen, Gert Pflugfelder, Uwe Walldorf, Isabel Guererro and the Developmental Studies Hybridoma Bank (DSHB) for antibodies; Steve Kerridge, Steve Cohen, Kwang-Wook Choi and Cheng-ting Chien for critical comments on the manuscript; Chiou-Yang Tang for fly stocks; and Chun-lan Hsu and Lili Chen for preparing the fly media. This study and M. K.-S. were supported by the National Science Council of the Republic of China (NSC-88-2312-B-001-016). A. S. was supported by a postdoctoral fellowship from Academia Sinica, Republic of China.