An interactive network of zinc-finger proteins contributes to regionalization of the Drosophila embryo and establishes the domains of HOM-C protein function

Though much is known about HOM-C/HOX control of development(McGinnis and Krumlauf, 1992; Akam, 1995; Biggin and McGinnis, 1997; Mann and Morata, 2000), many unanswered questions remain. Arguably, the most important is how different HOM-C proteins activate or modulate different target genes. In vitro, all HOM-C proteins bind to similar, relatively simple DNA sequences(Hoey and Levine, 1988; Ekker et al., 1994; Walter et al., 1994; Biggin and McGinnis, 1997), and there is evidence that this may be true in vivo as well(Walter et al., 1994; Carr and Biggin, 1999; Li et al., 1999). Surrounding bases can influence binding strength, but there appears to be little specificity or, more appropriately, selectivity, in the DNA-binding properties of different HOM-C proteins. Interactions with co-factors provide the likely resolution of this dilemma, but, currently, few co-factors are known (e.g. Peifer and Wieschaus, 1990; Röder et al., 1992; Chan et al., 1994; Castelli-Gair, 1998; Mann and Morata, 2000; Mahaffey et al., 2001).

Previously, we provided genetic evidence that the C2H2 zinc-finger proteins encoded by disconnected (disco) and disco-related(disco-r) are redundant co-factors for the gnathal HOM-C proteins,Deformed (DFD) and Sex Combs Reduced (SCR)(Mahaffey et al., 2001). DFD and SCR are required during development of the Drosophila larval gnathal (mandibular, maxillary and labial) segments. Embryos lacking disco and disco-r develop with a phenotype similar to those lacking these HOM-C genes, and this phenotype is due, at least in part, to reduced expression of DFD and SCR target genes. As the gnathal HOM-C proteins are not required for disco and disco-r activation,and vice versa, we proposed that these redundant proteins were potential co-factors required for DFD and SCR function.

Many questions remain concerning this proposal. For example, are disco and disco-r required for all DFD functions, and do they have patterning roles independent of the HOM-C proteins? Several studies have shown that ectopic DFD can induce maxillary structures in the trunk segments. Does this indicate that DISCO and DISCO-R are required only for DFD function in the gnathal segments? Interestingly, there are several similarities between disco and disco-r and the trunk-specification gene teashirt (tsh). Each encodes a zinc-finger transcription factor and functions as a genetic co-factor during HOM-C specification of segment identity. The DISCO proteins and TSH are required in multiple segments where they interact with different HOM-C proteins. Expression of the disco genes and tsh abut at the gnathal-trunk boundary, possibly suggesting similar roles, but in different regions of the embryo. Here, we address these issues by examining the role of DISCO, with and without DFD, and the interplay between DISCO and TSH. We conclude the following: (1) alone, DISCO appears to impart a gnathal segment type; (2) cells can respond to the gnathal HOM-C protein DFD only where DISCO is present; and (3) TSH represses disco (and disco-r)expression in the trunk, thereby preventing gnathal traits from developing in the trunk segments. These observations lead us to propose a new model for the specification of segment identity within the Drosophila embryo.

Flies were reared on standard cornmeal-agar-molasses medium. The paired-Gal4 (prd-Gal4) stock was obtained from Dr A. Bejsovec (Duke University), the UAS-Dfd fly stock from Dr T. Kaufman (Indiana University), the UAS-tsh-13 and tsh⁸ lines from Dr S. Kerridge (CNRS Marseille France),and the armadillo-Gal4 (arm-Gal4) line from Dr W. McGinnis (University of California, San Diego). The UAS-Scr fly stock was obtained from the Bloomington, Indiana Drosophila Stock Center.

We induced ectopic expression, at 25°C, using arm-Gal4(Sanson et al., 1996) and prd-Gal4 (Yoffe et al.,1995) drivers with analogous results (referred to as arm→disco and prd→discorespectively, below).

Embryos were collected and prepared for cuticle examination following procedures described previously (Pederson et al., 1996). Females were allowed to lay eggs for up to 24 hours, and embryos were aged for at least 24 hours before fixing the unhatched terminal larvae.

To obtain flies expressing Dfd in the trunk segments of embryos lacking disco and disco-r, we crossed Df(1)XR14/FM7c females to UAS-Dfd (II) homozygous males. The non-FM7c female progeny were crossed to homozygous prd-Gal4 males producing males hemizygous for Df(1)XR14 and lacking disco and disco-r. We could recognize those ectopically expressing Dfd, as ectopic DFD disrupts anterior head development, thereby further aggravating the phenotype of the Df(1)XR14 hemizygotes.

Localization of mRNA and proteins followed the protocols essentially as described previously (Pederson et al.,1996). Probes for disco and disco-r mRNAs were from Mahaffey et al. (Mahaffey et al.,2001). For other mRNA localizations, probe templates were obtained from Drosophila genomic DNA using PCR. The primers used to generate clones were as follows: pannier, ACATTACGGACAGGCGACAC (forward) and TGCAAACAAGGCCGAGTAG (reverse); salm, GCATACCAGAGCAAAGCACA (forward)and GATAACCGCGGCACCCGATCACAGACCA (reverse); tsh, GCGTACCTGCACATGGTGGC(forward) and GATCTCCGCGGCTGACTCTCGGCAGG (reverse).

In otherwise normal embryos, ectopic DFD induces cirri and, occasionally,sclerotized mouthpart-like material in the trunk segments(Kuziora and McGinnis, 1988; Gonzalez-Reyes et al., 1992). These DFD-induced structures develop at or near regions of discoexpression, near the Keilin's Organ primordia in the thorax and in analogous positions in the abdominal segments (Fig. 1G, Fig. 3A,D). We suspected this endogenous DISCO was supporting development of the ectopic maxillary structures, and therefore these structures would disappear if embryos lacked disco and disco-r. To test this, we ectopically expressed Dfd in embryos hemizygous for Df(1)XR14, and, as expected, found no evidence of ectopic maxillary structures (Fig. 1A,B). We conclude that DFD could not transform the trunk segments toward maxillary identity in the absence of disco and disco-r.

If, indeed, DFD and DISCO are required for maxillary development, ectopic co-expression should activate maxillary-specific target genes in the trunk segments. We examined the expression of several DFD target genes, Distal-less, Serrate, reaper and 1.28. Though all appear to be regulated jointly by DFD and DISCO, we present results with 1.28,as its expression is less complex, and it is a useful marker for maxillary identity (Mahaffey et al.,1993; Mohler et al.,1995; Pederson et al.,1996; Pederson et al.,2000; Mahaffey et al.,2001). We show results using the prd-Gal4 driver, as this allowed comparison of normal and manipulated segments within the same embryo,but comparable results were obtained with arm-Gal4. The pattern of prd-Gal4 distribution in an early stage 12 embryo is shown in Fig. 1H and diagramed in Fig. 1I.

Alone, prd→disco had no effect on 1.28transcript distribution (Fig. 1D); transcripts accumulated as in wild type embryos(Fig. 1C). prd→Dfd, however, caused significant accumulation of 1.28 transcripts in the posterior labial epidermis(Fig. 1E), and we noted slight accumulation in a few cells near the posterior edge of the prd→Dfd segments (not visible in the image). That DFD induced 1.28 expression in the labial segment was expected as disco is normally expressed in many labial cells(Lee et al., 1991; Mahaffey et al., 2001) and ectopic DFD transforms the labial segment toward a maxillary identity(Kuziora and McGinnis, 1988). The weak expression in the trunk segments was unexpected, but was explained by the fact that ectopic DFD activated disco (see below). Co-expression of disco and Dfd caused significant accumulation of 1.28 transcripts in the posterior epidermis of every other trunk segment (Fig. 1F) overlapping with prd-Gal4 expression. We conclude from these experiments that the presence of DISCO makes the trunk segments competent to activate DFD target genes, and allows ectopic DFD to function in the presence of the trunk specification system.

The weak expression of 1.28 in prd→Dfdembryos did not coincide with disco expression in the Keilin's Organ precursors, but was more lateral and posterior. Because all of our other results indicated that DFD and DISCO are required together, we examined disco expression in prd→Dfd and arm→Dfd embryos. In both cases, ectopic DFD activated disco (Fig. 3A,C);this induction is likely to be responsible for the low level of 1.28RNA accumulation in UAS-Dfd embryos. Our previous results indicate that DFD is not required for disco expression in the gnathal segments(Mahaffey et al., 2001), so we suspect this DFD induction of disco reflects that DFD can modulate disco expression.

Because, in an otherwise normal embryo, ectopic DFD causes only a limited trunk to maxillary transformation, and as disco is required for this,we suspected that co-expression of disco and Dfd should yield a more complete transformation. Surprisingly, this was not the case. On average, more cirri developed in the trunk segments upon co-activation with arm-Gal4, but not with prd-Gal4 (data not shown). This suggested that DISCO and DFD were not sufficient to induce a stronger transformation; either something else was needed, or the trunk to maxillary transformation was inhibited in a manner that could not be overcome by additional DISCO. Components of the trunk specification program, for example TSH (Fasano et al., 1991; Röder et al., 1992), are likely inhibitors, so we examined the effect of ectopic DFD and DISCO in tsh mutant embryos.

Röder et al. (Röder et al.,1992) reported that the trunk segments are partially transformed toward head identity in embryos lacking TSH, as indicated by ectopic sclerotic material in the trunk segments, and changes in the trunk peripheral nervous system. Though we occasionally observed small patches of sclerotic material in the cuticle of homozygous tsh⁸ mutant embryos, we never observed mouth hook-like structures (Fig. 2B). Surprisingly, ectopic DFD caused a stronger transformation when embryos lacked TSH (Fig. 2D) than in otherwise normal embryos(Fig. 2C). Cirri and sclerotized material appeared in most Dfd-expressing segments, and the sclerotized material more closely resembled normal maxillary mouth hooks. Co-expression of disco and Dfd in embryos lacking TSH produced an even more consistent transformation(Fig. 2E,F), where nearly every expressing segment produced cirri and well-formed mouth hooks. We note that no mouthpart structures were produced in embryos lacking disco and disco-r and tsh, regardless of whether or not ectopic DFD was present (data not shown).

These results indicated that TSH hindered the DFD-induced trunk to maxillary transformation, but raised the question why was DFD sufficient to cause a more complete transformation when TSH was absent? As discowas still required for the transformation, and because absence of tshcauses a partial trunk to head transformation, it seemed likely that TSH may repress head specifying genes, genes such as disco. Therefore, we examined disco (and disco-r with analogous results) mRNA distribution in embryos lacking tsh and found it was more widely distributed in these embryos (Fig. 3B). Normally, in the trunk, disco mRNA accumulates in the Keilin's Organ primordia and in analogous positions in the abdominal segments (Fig. 3A,D). In tsh mutants, the Keilin's Organ primordia were absent, and disco mRNA was broadly distributed in the ventral and ventrolateral portion of trunk segments. Ectopic disco mRNA did not extend into the dorsal trunk epidermis, where absence of TSH has little or no effect(Röder et al., 1992). Interestingly, disco mRNA distribution in tsh mutants was very similar to that observed in arm→Dfd embryos (see above), with one notable exception. disco mRNA was still present in the Keilin's Organ primordia of embryos ectopically expressing Dfd,but not in tsh mutants (compare Fig. 3B, tsh mutant,with Fig. 3C, ectopic Dfd).

To further test the repression of disco by TSH, we ubiquitously expressed tsh using the arm-Gal4 driver so that TSH would accumulate in all of the gnathal cells. As shown in Fig. 3E,F, TSH altered normal gnathal expression of disco. At the beginning of germband retraction,correlating with the onset of arm-Gal4 expression, discomRNA levels decreased (Fig. 3E)until, by the end of germband retraction, the normal gnathal distribution was no longer detectable (Fig. 3F). Interestingly, disco mRNA was not completely eliminated. In each gnathal lobe, disco mRNA accumulated in a small cluster of cells(Fig. 3F) resembling that observed in the thoracic Keilin's Organ precursors(Fig. 3A,D); indeed, ectopic TSH transforms the labial sense organ into one resembling a Keilin's Organ(de Zulueta et al., 1994).

We conclude from the above observations that DISCO and DFD can override the trunk specification system to generate maxillary identity. One manner in which this could occur is for our manipulations to repress expression of the trunk specification system. We noted that ectopic disco expression did not repress tsh transcription, in contrast to the reverse described above(data not shown). Still, repression could occur through the trunk HOM-C genes, and in this regard, it is worth noting that lack of the trunk HOM-C input does give rise to sclerotized material in the trunk segments(Struhl, 1983, Sato et al., 1985). However,our manipulations did not alter the normal distribution of trunk HOM-C proteins (Fig. 4). We examined the distribution of several trunk HOM-C proteins in embryos of all manipulations used in this study, using the arm-Gal4 driver to have the broadest possible effect. We found no indication that HOM-C protein accumulation was significantly altered, other than because of the grossly aberrant morphology of later embryos ectopically expressing DISCO. Even then,HOM-C proteins were distributed in the proper register (data not shown). We also examined Labial distribution, as embryos lacking tsh were reported to accumulate Labial in small clusters of cells in the trunk. However, we could not detect ectopic labial expression in our tsh⁸ embryos. We conclude that the combination of DISCO and DFD can override the trunk identity system, redirecting development toward maxillary identity. Clearly, this was more complete when the trunk specification system is compromised, as it is when TSH is absent.

That disco and disco-r are ectopically expressed in tsh mutants could explain the trunk to head transformation reported in these embryos (Röder et al.,1992). Considering this, we re-examined the effects of ectopic DISCO to see if this would override the trunk specification system,transforming the trunk to a head identity. In embryos ubiquitously expressing disco, germband contraction fails and a hole appears in the dorsal epidermis, indicating that DISCO may interfere with dorsal closure(Robertson et al., 2002). We stained prd→disco embryos with anti-Engrailed/Invected(EN) antibodies (DiNardo et al.,1985) to monitor the fate of cells in the posterior compartments of the trunk segments, as these cells, and a few cells anterior to these,would be expressing disco. In normal stage 13 embryos EN accumulates in dorsoventral stripes about two cells wide marking the posterior compartment of each segment (Fig. 5A). The EN-positive cells can be followed during dorsal closure, when the cells of the trunk segments extend toward the dorsal midline and fuse with cells from the contralateral side. In the affected segments of prd→disco embryos, the EN-expressing cells and a few cells anterior to these did not extend toward the dorsal midline(Fig. 5B). Only those cells in the anterior half of the affected segments - those cells not expressing prd→disco - completed dorsal closure. We noted that the lack of dorsal closure caused the altered trunk segments to acquire a shape similar to the gnathal lobes.

pannier (pnr) encodes a GATA class zinc-finger protein required for dorsal closure (Herranz and Morata, 2001). In early stage 12 embryos, pnr mRNA accumulates along the dorsal edge of the segments, from the posterior maxillary to the eighth abdominal segment(Fig. 5C). In prd→disco embryos, this continuous line of pnrmRNA accumulation was broken (Fig. 5D). Double-labeling with EN antibodies confirmed that the gaps in pnr mRNA accumulation coincided with the cells expressing prd→disco (data not shown). Repression of pnrwas transient. Later in development, as ectopic disco mRNA faded(late stage 12 to early 13), pnr mRNA was detected in the dorsal limits of the affected segments (data not shown), but apparently, this was too late to rescue dorsal closure. At this time, we do not know whether repression of pnr is direct.

Dorsal development is very limited in the gnathal segments, where disco is normally expressed. The dorsal ridge is a reduced segment-like structure derived from the gnathal segments(Fig. 5E), and it is the anteriormost structure able to adopt a dorsal fate(Rogers and Kaufman, 1996). Many of the cells that will give rise to the dorsal ridge appear as de novo EN-expressing cells along the dorsal edge of the maxillary and labial lobes(Rogers and Kaufman, 1996). Though disco is expressed in many gnathal cells(Lee et al., 1991; Mahaffey et al., 2001), it is not expressed in these dorsal ridge precursors. In fact, the dorsal ridge was quite reduced or eliminated when disco was ectopically expressed in these cells (Fig. 5F). This prompted us to ask whether dorsal ridge development was altered in embryos lacking disco and disco-r, and, indeed, this appeared to be the case. In male embryos carrying Df(1)XR14, the dorsal ridge was enlarged and joined with the labial, and sometimes maxillary, lobes(Fig. 5G). We conclude that normal disco expression is needed to limit gnathal contribution to the dorsal ridge.

Ectopic disco expression disrupted other aspects of trunk development. Previously, we showed that DISCO repressed denticle formation(Robertson et al., 2002), and now we find that other aspects of trunk development are also disrupted. The dorsal trachea and oenocytes were absent(Fig. 6A,B), as indicated by lack of spalt-major (salm) expression, which is required for formation of these structures(Kühnlein et al., 1994). We note that other regions of salm expression were unaffected. The trunk peripheral nervous system was also altered by ectopic discoexpression. Visualized using anti-22c10/Futsch antibodies(Hummel et al., 2000), there is a characteristic pattern of sensory neurons produced in each trunk segment(Campos-Ortega and Hartenstein,1997), and ectopic DISCO altered these in several ways(Fig. 6C,D). The chordotonal organs were absent as were other sensory structures. Ectopic DISCO did not simply eliminate neural structures. Sensory neurons formed, but they did not resemble those normally found in the trunk. We are uncertain of their identity, but suggest that they have a mixed gnathal/trunk identity as both DISCO and TSH are present in these segments. Unknowingly, the role of DISCO in the absence of TSH has been examined previously, while examining tshmutants. As we described above, disco and disco-r are activated in the trunk of embryos lacking TSH, and Röder et al.(Röder et al., 1992)concluded that the trunk neurons can acquire a gnathal identity in these embryos.

Though our results above deal only with maxillary identity, our prior genetic analysis indicated that disco and disco-r were required for labial development, as well. To determine whether or not this was a general mechanism governing development throughout the gnathal segments, we examined the role of disco with Scr, using proboscipedia (pb) as a marker, which has been shown to be a target of SCR in the labial segment (Rusch and Kaufman, 2000). PB ectopically accumulates in the first thoracic segment (T1) of tsh mutant embryos(Rusch and Kaufman, 2000), so it seemed likely that this was due to the presence of SCR in T1 and the de-repression of disco and disco-r in these embryos. To test this, we co-expressed disco and Scr in the trunk segments using the prd-Gal4 driver, and indeed, this leads to significant ectopic accumulation of PB, when compared to ectopic Scr alone(Fig. 7). This suggests that DISCO has a similar role in maxillary and labial development. We note that expression of pb was somewhat spatially limited. This could be due to the use of the prd-Gal4 driver, the altered morphology of the affected segments, or perhaps other factors limit pb expression. A similar enhancement of PB accumulation did not occur with DISCO and DFD.

disco was initially identified in a screen for mutations affecting neural development (Steller et al.,1987). It was not until the discovery of disco-r that a patterning role was uncovered (Mahaffey et al., 2001). The phenotype of terminal embryos lacking disco and disco-r is similar to those lacking the gnathal HOM-C genes Dfd and Scr; that is, structures from the gnathal segments (mandibular, maxillary and labial) are missing. This phenotype is due to reduced expression of DFD and SCR target genes. As HOM-C protein distribution is normal in disco, disco-r null embryos, and vice versa, these factors appear to act in parallel pathways.

We have extended these studies and show that: (1) DFD can only direct maxillary developmental when DISCO and/or DISCO-R are present; (2) TSH represses disco (and disco-r), helping to distinguish between trunk and gnathal segment types, and thereby establishing domains for appropriate HOM-C protein function; and (3) when ectopically expressed in the trunk, DISCO represses trunk development and may transform these segments towards a gnathal segment type.

Though HOM-C genes have a clear role in establishing segment identities, ectopic expression often has only a limited effect. Our data indicate that, for DFD, this restriction arises because of the limited distribution of DISCO in the trunk segments. There are two important conclusions from these observations. First, the spatial distribution of DISCO establishes where cells can respond to DFD, and this is probably true for SCR as well. Cells expressing disco develop a maxillary identity when provided with DFD, even though this may not have been their original HOM-C-specified fate. This highlights the second point: the combination of DISCO and DFD overrides normal trunk patterning, without altering expression of tsh and trunk HOM-C genes. As with the maxillary segment,identity is lost in the mandibular and labial segments when embryos lack disco and disco-r. This indicates that DISCO and DISCO-R may have similar roles in all gnathal segments. That co-expression of DISCO and SCR in the trunk activates the SCR gnathal target gene pb strengthens this conclusion. Therefore, we propose that DISCO defines the gnathal region,and establishes where the gnathal HOM-C proteins DFD and SCR can function.

Alone, ectopic DISCO significantly alters development, indicating that DISCO has a morphogenetic ability, separate from gnathal HOM-C input. As DISCO is required for normal gnathal development, we suspect that discospecifies a general gnathal segment type. Definitive identification is difficult because of the lack of morphological or molecular markers that denote a general gnathal segment type. Yet, there is support for the conclusion that disco expression establishes a gnathal segment type. Ectopic DISCO can, to some extent, override the trunk specification system and repress trunk development (repressing denticles, oenocytes and trachea). Furthermore, ectopic DISCO blocks dorsal closure, which is similar to the role of endogenous DISCO in the gnathal segments.

Perhaps the most compelling evidence that DISCO specifies a gnathal segment type comes from the observation that disco is activated in the trunk segments when embryos lack TSH. The identity of the trunk segments in tsh mutant embryos is somewhat uncertain. Fasano et al.(Fasano et al., 1991) and Röder et al. (Röder et al.,1992) suggested that some aspects of the tshphenotype indicate the trunk segments acquire gnathal characteristics; for example, the ventral neural clusters appear to be transformed to a gnathal-like identity (as mentioned above). Röder et al. state that`Mutations in the tsh gene can therefore be interpreted in two ways;either they partially transform the trunk segments into a gnathal-like identity, and in particular the prothoracic segment into a labial one, or they cause a non-specific change in segmental identity perhaps due to cell death';however, they also report that the loss of tsh and the trunk HOM-C genes may transform the trunk cuticle toward anterior head cuticle. Again, the difficulty in assigning an identity is due to the lack of a readily discernable gnathal morphological or molecular marker. We present evidence that disco and disco-r are reliable molecular markers for gnathal identity, and we show that disco mRNA is present in the ventral and lateral regions of the trunk segments in tshmutant embryos. This expression of disco coincides, spatially, with the region of the trunk that is transformed in tsh mutant embryos. UAS-driven disco does mimic some aspects of tsh mutants,denticles are reduced and the ventral chordotonal neurons do not develop, but as TSH is still present, the transformation caused by ectopic discomay be incomplete. Finally, DFD cannot induce maxillary structures, even in tsh mutants, when disco and disco-r are absent. This reinforces the role for DISCO in establishing gnathal identity, and indicates that the ectopic DISCO present in embryos lacking TSH is functional. Therefore, considering these arguments, we propose that DISCO and DISCO-R establish the gnathal region of the Drosophila embryo, and in this regard, they function similarly to TSH, which specifies the trunk region.

There are other parallels between DISCO/DISCO-R and TSH. They are regionally expressed zinc-finger transcription factors, and they are required in parallel with the HOM-C proteins for proper segment identity. Furthermore,the distribution of these proteins establishes domains in which specific HOM-C proteins can properly direct embryonic development. Our data reveal a regulatory relationship between TSH and disco (and disco-r),indicating they are part of an interacting network that helps regionalize the Drosophila embryo. The HOM-C proteins then establish specific segmental identities in the appropriate region. A schematic of this model is presented in Fig. 8. In the trunk segments, TSH, along with the trunk HOM-C proteins, specifies the trunk segment characteristics, in part by repressing disco and, thereby,preventing gnathal characteristics from arising in the trunk segments. Our model requires that tsh expression be limited to the trunk segments,and we propose this is accomplished by another C2H2 zinc-finger protein, SALM. Röder et al. (Röder et al.,1992) demonstrated that tsh expression expands into the posterior gnathal and posterior abdominal segments in embryos lacking SALM. Therefore, SALM establishes the boundary between the TSH and DISCO domains. We stress that, at this time, we do not know what parts of this regulation are direct. Interestingly, other zinc-finger transcription factors are responsible for positioning salm expression (Kühnlein et al., 1997), so that a more extensive hierarchy of zinc-finger transcription factors leads to regionalization, eventually establishing the domains of HOM-C protein function. We also note that TSH has other roles than just repressing disco. TSH actively establishes the trunk region, just as DISCO does the gnathal. It is also noteworthy that ectopic TSH activated discoin the labial sense organ primordia, leading to a Keilin's Organs fate, as occurs in the thoracic segments. Therefore, for unknown reasons, TSH changes from a repressor of disco to an activator in these cells. This observation highlights the complex interplay between factors like TSH and DISCO, and it will be interesting to determine what causes these opposing roles.

Many other questions remain. For example, how are the expression domains for these factors established? It is clear that SALM could form a boundary separating gnathal from trunk, but in salm mutants, tsh is only ectopically activated in the posterior labial segment, not in every gnathal segment (Röder et al.,1992). This implies that SALM forms a boundary, not by repressing tsh throughout the head, but by, in a sense, drawing a line between the head and trunk regions. What then prevents tsh expression from crossing that line and extending further into the gnathal segments in salm mutants? Is there an activator of tsh that is limiting,another gnathal repressor, or is something else involved? Likewise, what activates tsh and disco? It is unlikely that lack of TSH is the only requirement for disco expression. More likely, this relies on the prior segmentation pathway. With regard to the HOM-C specification of segment identity, questions remain as to how the zinc-finger proteins establish where specific HOM-C proteins can function. Are the zinc-finger proteins co-factors or simply a parallel pathway? Furthermore, if they are co-factors for the HOM-C proteins, how can different HOM-C proteins establish different segment identities with the same co-factor (for example, DFD and SCR with DISCO), or how can different co-factors alter the role of a HOM-C protein(SCR with DISCO or TSH)?

Finally, we are left with the question of whether or not factors such as DISCO and TSH establish head/trunk domains and delimit HOM-C protein function only in the Drosophila embryo, in all stages of Drosophilaor in other animals as well. Though this remains to be tested experimentally,there are indications that this may be a general mechanism. Homologues of these zinc-finger genes are found in vertebrates and in other invertebrates(Caubit et al., 2000; Knight and Shimeld, 2001),and, although only limited data are currently available(Caubit et al., 2000) (M. K. Patel and J.W.M., unpublished), expression data indicate that these genes may have similar roles to their Drosophila counterparts during embryonic patterning. In an informative experiment by Brown et al.(Brown et al., 1999), they expressed the Tribolium Dfd homologue, Tc-Dfd, in Drosophila embryos lacking the endogenous Dfd gene and showed that persistent expression of Tc-Dfd could rescue maxillary development. Though, at present, it is not known whether or not a direct interaction is required between DISCO and DFD, this result would indicate that the Tribolium DFD protein can fulfill the same roles as the Drosophila protein, and, therefore, it must be able to function with the Drosophila regionalization system. In any case, it will be important to investigate and interpret the role of the regionalizing genes as they relate to development and evolution of body pattern in other animals, and to ask whether a similar network is involved in patterning all animals.

We thank Dr A. Bejsovec (Duke University) for the paired-Gal4 stock, Dr P. Estes (North Carolina State University) for the 22c10 antibodies, Dr T. Kaufman (Indiana University) for the UAS-Dfd fly stock and the PB antisera, Dr W. McGinnis (University of California, San Diego) for the armadillo-Gal4 lines, and Dr S. Kerridge (CNRS Marseille France) for the tsh⁸mutants and the UAS-tsh-13 fly line. This research was supported by a grant (IBN0090440) from the Developmental Mechanisms Program of the National Science Foundation to J.W.M. L.K.R. was partially supported by NIH pre-doctoral fellowship GM-08443-10.