Abstract
The specific functions of gene products frequently depend on the developmental context in which they are expressed. Thus, studies on gene function will benefit from systems that allow for manipulation of gene expression within model systems where the developmental context is well defined. Here we describe a system that allows for genetically controlled overexpression of any gene of interest under normal physiological conditions in the early Drosophila embryo. This regulated expression is achieved through the use of Drosophila lines that express a maternal mRNA for the yeast transcription factor GAL4. Embryos derived from females that express GAL4maternally activate GAL4-dependent UAS transgenes at uniform levels throughout the embryo during the blastoderm stage of embryogenesis. The expression levels can be quantitatively manipulated through the use of lines that have different levels of maternal GAL4 activity. Specific phenotypes are produced by expression of a number of different developmental regulators with this system, including genes that normally do not function during Drosophila embryogenesis. Analysis of the response to overexpression of runt provides evidence that this pair-rule segmentation gene has a direct role in repressing transcription of the segment-polarity gene engrailed. The maternal GAL4 system will have applications both for the measurement of gene activity in reverse genetic experiments as well as for the identification of genetic factors that have quantitative effects on gene function in vivo.
THE Drosophila embryo provides a remarkable demonstration of the productive interplay between developmental genetics and mechanistic studies on gene function. For example, the central role of transcriptional regulation in patterning the embryo resulted from studies on genes identified through the pioneering mutational studies of Nüsslein-Volhard and Wieschaus (1980). These studies have generated a solid framework for functional studies on the transcription factors that participate in these processes. Indeed, experiments with the Drosophila embryo provide several of the most elegant and well-understood examples of transcriptional regulation in developmental biology. Although the roles of many genes in the embryo were initially deduced from loss-of-function phenotypes, studies on gene function have also benefited from analysis of gain-of-function phenotypes produced by overexpression. For example, numerous investigations on the regulatory interactions and mechanisms of transcriptional regulation that are involved in the segmentation pathway have used the Drosophila heat-shock promoter to induce ectopic gene expression (Struhl 1985; Ish-Horowicz and Pinchin 1987; Morrisseyet al. 1991; Fitzpatricket al. 1992; Manoukian and Krause 1992, 1993; Tsai and Gergen 1994, 1995; Johnet al. 1995; Aronsonet al. 1997; Donget al. 1998). One concern in the interpretation of all of these experiments is the fact that the regulatory responses may be nonphysiological as they are obtained in embryos that have been heat-shocked.
We have been using the Drosophila embryo as a model to investigate the function of the pair-rule segmentation gene runt. Runt is the founding member of the Runt domain family of transcriptional regulators (Kagoshimaet al. 1993) and has pivotal roles in sex determination, segmentation, and neurogenesis in the Drosophila embryo (Gergen and Wieschaus 1986; Duffy and Gergen 1991; Duffyet al. 1991; Torres and Sanchez 1992; Dormandand Brand 1998; Krameret al. 1999). Lozenge, a second Runt domain protein in Drosophila, has postembryonic roles in patterning in the antenna and eye (Dagaet al.1996; Floreset al. 1998; Guptaet al. 1998). In vertebrates, Runt domain genes have important roles in the normal development of blood and bone, and mutations in these genes are associated with defects in these processes in humans (Okudaet al. 1996; Wanget al. 1996; Ducyet al. 1997; Komoriet al. 1997; Ottoet al. 1997; Traceyet al. 1998).
In Drosophila, runt is required for the proper transcriptional regulation of a number of different genes during early embryogenesis. In some instances runt functions to activate transcription (Tsai and Gergen 1995; Krameret al. 1999), whereas on other targets runt plays a role in transcriptional repression (Manoukian and Krause 1993; Tsai and Gergen1994). Recent studies using heat-shock expression assays have further revealed two independent modes of transcriptional repression by the Runt protein. The repression of the pair-rule genes even-skipped (eve) and hairy involves recruitment of the corepressor protein Groucho through a C-terminal VWRPY motif that is conserved on other Runt domain proteins (Aronsonet al. 1997). In contrast, repression of the segment polarity gene engrailed (en) and the head gap gene orthodenticle (otd) is achieved through a VWRPY- and Groucho-independent pathway (Aronsonet al. 1997; Tsaiet al. 1998). Although the mechanisms that account for the various regulatory activities of the Runt protein are not understood, our working model is that Runt activity is modulated through specific interactions with other context-dependent transcriptional regulators.
The modular GAL4 system is an alternative and important tool that has been developed for the manipulation of gene expression in Drosophila (Brand and Perrimon 1993). This system takes advantage of the ability of the yeast transcriptional activator GAL4 to function in Drosophila cells (Fischeret al. 1988). Drosophila strains that express GAL4 specifically at one stage of development or in a specific cell type can be mated to strains carrying a GAL4-responsive transgene that expresses any gene of interest (UASG-X) in order to induce expression of gene X in a genetically controlled manner. A number of investigators have used an enhancer-trap version of a GAL4-expressing P-element transposon to generate a large number of lines that are useful in different developmental contexts (Brand and Perrimon 1993; Yanget al. 1995; Gustafson and Boulianne 1996; Manseauet al. 1997). Notable by their absence in these collections are lines that express GAL4 at significant levels during the early stages of embryogenesis, the stages that would be most useful for studies on segmentation gene products.
Here we describe a system that allows for GAL4-driven ectopic expression in the early embryo. We constructed a GAL4 expression vector that uses the promoter from the maternally expressed gene nanos and a 3′ untranslated (UTR) region from a tubulin mRNA to drive maternally dependent uniform activation of GAL4-dependent transgenes during the blastoderm stage of embryogenesis. Through the use of NGT (nanos–GAL4–tubulin) lines that have different levels of GAL4 activity, it is possible to quantitatively manipulate expression in the embryo over a range that extends up to an estimated 125,000 molecules/blastoderm cell. Specific phenotypes are produced by expression of different developmental regulators with this system, including genes that normally do not function during embryogenesis. We further characterized the use of this system for the analysis of runt function. Importantly, all of the targets identified as being subject to regulation by runt using the heat-shock assay system also respond when runtexpression is driven with the maternal GAL4 system. However, experiments with the GAL4 system reveal clear quantitative differences in the sensitivity of different targets. The potential applications of this system for quantitative studies of gene function in the Drosophila embryo are discussed.
MATERIALS AND METHODS
Plasmid construction: A CaSpeR P-element transformation vector was constructed that contains the coding region of yeast GAL4 immediately downstream of the nanospromoter and upstream of DNA containing the 3′ UTR of the Drosophila αTub84B gene. The nanos promoter and ~250 bp of 5′ UTR are contained in a 1.1-kb HindIII + NdeI restriction fragment that was isolated from plasmid pBS-P[nos], generously provided by L. Gavis (Princeton University). One step of the cloning strategy included destruction of an initiator methionine for the Nanos protein that is encoded at the NdeI site of this fragment. The GAL4-encoding fragment was isolated as a HindIII fragment from plasmid pGAT-B (Brand and Perrimon 1993). The αTub84B 3′ UTR was contained within an 850-bp XhoI + HindIII fragment from plasmid pTα1-5′-3′ (Theurkaufet al. 1986). These fragments were inserted into the BamHI site of pCaSpeR (Thummelet al. 1988). The NGT fusion gene is oriented within this vector such that it is transcribed from the opposite DNA strand as the white marker gene in a divergent manner. Further details of the cloning strategy used to generate this plasmid are available on request.
A UAS-runt construct was made by cloning a 3.2-kb BamHI fragment isolated from pCaSpeR:hs-runt (Tsai and Gergen 1994) into the BglII site of the pUAS-T vector (Brand and Perrimon 1993). The portion of the runt mRNA contained within this segment begins four nucleotides upstream of the initiator methionine and extends through the polyadenylation signal to the poly(A) tail of a full-length runt cDNA.
Fly strains and crosses: Drosophila strains were maintained on standard cornmeal/yeast/sugar and agar media. General information on marker mutations and balancer chromosomes is available through FlyBase (1999). Germ-line transformants carrying the P{w[+mC] ScerGAL4[nos.PG] = GAL4-nos.NGT} and P{UAS-runt.T} transposons were recovered by standard P-element-mediated germ-line transformation protocols using the p:Δ2-3 helper plasmid. The P{UAS-runt.T}232 line was recovered as 1 of 14 initial independent lines and was retained as a representative strong line based on experiments with a Kr-GAL4 driver (M. Klingler and J. P. Gergen, unpublished results). Preliminary characterization based on the lethality obtained with various UAStransgenes indicated that 2 of the initial NGT lines, P{GAL4-nos.NGT}9 and P{GAL4-nos.NGT}11, had the highest levels of activity. These 2 lines were mapped to the third and second chromosomes, respectively, and stocks homozygous for each of these chromosomes were established. P-transposition mediated by the Δ2-3 third chromosome (Robertsonet al. 1988) was used to generate hops from the original P{GAL4-nos.NGT}11 line. All of the lines reported here were obtained as hops on chromosome II that are associated with a change in eye color. Experiments to generate further NGTlines on both the second and third chromosomes, as well as on different balancer chromosomes, are in progress.
Flies carrying an X-chromosome-linked transposon for GAL4-dependent expression of lozenge (lz) were obtained from Utpal Banerjee (UCLA). The UAS transgenes for expression of lacZ (P{UAS-lacZ.B}4-1-2), decapentaplegic (dpp, P{UAS-dpp.S} 42B.4), en(P{UAS-en.Y}4-1), and the different isoforms of pointed (pnt[P1], P{UAS-pnt.P1}3), and pnt[P2], P{UAS-pnt.P2}2) were obtained from the Bloomington stock center. The transposon for the P{UAS-runt.T}232 line maps on chromosome II and is homozygous viable. The P{UAS-runt.T}U15 line, which also maps on chromosome II, was generated by Δ2-3-mediated mobilization. Quantitation by RNase protection indicates that the U15line is expressed at approximately three-fold higher levels than the 232 line (Li 1999). The relative viability of flies carrying different UAS transgenes was determined by mating males heterozygous for the transgene and the appropriate balancer chromosome (either CyO or TM3) to virgin females from the pertinent NGT stock. In experiments with P{UAS-en.Y}4-1 (which is on the TM6 balancer), males were heterozygous for the ruPrica marker chromosome. These crosses were carried out in vials at 25° in uncrowded conditions.
Embryo manipulation: For cuticle preparations, nonhatching embryos were dechorionated in bleach, rinsed with distilled water, and mounted in a 1:1 mixture of lactic acid and Hoyer’s. In situ hybridization was carried out as described previously (Klingler and Gergen 1993). Digoxigenin-labeled (Boehringer Mannheim, Indianapolis) RNA riboprobes to detect the lacZ, runt, eve, and fushi tarazu (ftz) mRNA transcripts were synthesized as described previously (Tsai and Gergen 1994). Expression of en was detected using a probe synthesized from HindIII digested pB:en (gift of D. Ish-Horowicz, ICRF) template with T7 RNA polymerase.
The quantitative measurement of β-galactosidase activity was done using a luminescent substrate and the Galacto-Light Plus detection kit (Tropix, Bedford, MA). Single living embryos of the appropriate stage were identified by observation under oil (Halocarbon Products), transferred into a microfuge tube, and homogenized in 50 μl of lysis buffer (100 mm potassium phosphate, pH 7.8, 0.2% Triton X-100, 1 mm dithiothreitol, 100 μg/ml bovine serum albumin). One-fifth of this homogenate was diluted into 70 μl of reaction buffer (100 mm sodium phosphate, pH 8.0, 1 mm magnesium chloride) containing a 1:100 dilution of the Galacto-Light Plus substrate stock solution (Tropix) and incubated at room temperature for 30–60 min. Activity was measured after addition of 100 μl of Light Emission Accelerator (Tropix) containing 10 mm H2O2 on an Optocomp I luminometer (MGM Instruments, Inc.). The relative light units reported were all from readings taken over a 30-sec interval. This assay was calibrated by adding purified β-galactosidase (Sigma, St. Louis) into extracts of control embryos. The assay was linear over the full range tested, from 0.46 to 330 pg with an average measurement of 15 × 103 light units/pg of enzyme.
RESULTS
A maternal effect GAL4 mRNA: The two-component GAL4 system (Brand and Perrimon1993) has been extremely useful for targeting gene expression to a number of different tissues and a number of different stages during Drosophila development. Notable by their absence has been the availability of GAL4 lines that drive expression during oogenesis and the earliest stages of embryogenesis. It has been suggested that this is due to the developmental regulation of factors that specifically affect either the translation of the GAL4 mRNA or the activity of the GAL4 protein during these stages. Regulation at the level of mRNA translation has been demonstrated for a number of genes that are expressed during oogenesis (Al-Atiaet al. 1985; Gavis and Lehmann 1994; Salleset al. 1994; Kim-Haet al. 1995; Markussenet al. 1995). Furthermore, there is also substantial evidence indicating that maternally expressed mRNAs contain signals that mediate transport from their site of synthesis in the nurse cells to the developing oocyte (MacDonald and Struhl 1988; Ephrussi and Lehmann 1992). This transport is important if the maternal mRNA is to be available for translation during the early stages of embryogenesis. On the basis of these observations, we reasoned that GAL4 transgenes that contained appropriate cis-regulatory elements for transcription, translation, and mRNA transport during oogenesis might be active during oogenesis and early embryogenesis. The two elements that we used to test this hypothesis were the promoter from the maternal effect gene nos, and the 3′ UTR from the αTub84Bgene. nos is specifically expressed in germ cells with high expression levels in the ovary in germ-line cysts from region 2 of the germarium (Wanget al. 1994; Forbes and Lehmann1998). The αTub84B gene product is abundant during both oogenesis and early embryogenesis (Matthewset al. 1989), suggesting that the 3′ UTR of this mRNA would allow both for transport into the oocyte and for translation during early embryogenesis. The anticipated properties of the NGT fusion gene that we constructed are schematically presented in Figure 1.
A number of transgenic lines carrying the P{GAL4-nos.NGT} were obtained by P-element-mediated germline transformation. P-element remobilization was also used to obtain additional derivatives of these initial lines. Characterization of these lines by a number of criteria indicated that they did indeed express GAL4 maternally, and that there were differences in the level of expression in the different lines. We used P{GAL4-nos.NGT}40, a relatively strong line (see below), to investigate the activity of the NGT transgenes. In situ hybridization reveals that embryos derived from homozygous P{GAL4-nos.NGT}40females express a paternally inherited UAS-lacZ reporter gene during the syncytial blastoderm stage, prior to cellularization (Figure 2B). Expression is observed at uniform levels throughout the embryo with the exception that no lacZ mRNA accumulates within the pole cells (Figure 2C). Expression levels increase through the completion of cellularization, gastrulation, and the process of germband extension (Figure 2, D and E). Reciprocal matings between females that are homozygous for the UAS-lacZ transgene and homozygous P{GAL4-nos.NGT}40 males confirm that NGT-driven expression is strictly maternally dependent. Thus, the increasing accumulation of UAS-lacZ mRNA during the postblastoderm stages is due to perduring maternal GAL4 activity at either the mRNA and/or protein level.
A sensitive photometric assay for β-galactosidase activity was used to quantitate the levels of NGT-driven gene expression. This assay used a luminescent substrate that allows for detection of UAS-lacZ expression in aliquots of extracts prepared from single embryos (see materials and methods). To generate a developmental time course, we performed assays on single embryos that were staged based on the time at which they initiated gastrulation. This is a readily observed and rapid morphogenetic process that occurs immediately upon completion of cellularization. The activity detected in early gastrula stage embryos is >100-fold greater than the background levels detected in control crosses (Table 1). This enzyme activity depends on not only transcription of the P{UAS-lacZ.B} transgene, but also translation of the lacZ mRNA prior to the completion of cellularization. Calibration experiments indicate that the level of activity detected per embryo at this stage is equivalent to that of 67 pg of β-galactosidase. This corresponds to ~1.25 × 105 molecules of β-galactosidase in each of the 6000 blastoderm cells. As found for the mRNA, β-galactosidase enzyme levels continue to increase during the germband extension stages (Table 1). Decreases in enzyme activity levels, which require degradation of both the lacZ mRNA and protein, do not become apparent until several hours later, when the embryos are in the process of germband retraction.
Figure 1.
GAL4-dependent activation with a maternal mRNA. GAL4-dependent transcriptional regulation is achieved in early embryos that contain maternally expressed GAL4 mRNA. The nos promoter is used to drive expression of the GAL4mRNA specifically during oogenesis. Inclusion of the 3′ untranslated region of a maternally expressed tubulin mRNA allows for uniform deposition of the GAL4 mRNA in the developing oocyte. GAL4-dependent transgenes can be inherited either from the male in the cross (as shown) or from the female.
Figure 2.
Developmental time course of NGT-driven gene expression. Whole-mount in situ hybridization reveals the pattern of expression of the P{UAS-lacZ.B}4-1-2 transgene during different stages of Drosophila embryogenesis. All embryos are oriented anterior to the left, dorsal side up. No expression is detected in embryos prior to formation of the syncytial blastoderm (A). Expression of the lacZ mRNA becomes detectable prior to nuclear division cycle 14, as shown by the cycle 12/13 embryo in B. Cycle 14 embryos in the process of cellularization (C) show uniform expression, except in the presumptive germ cells at the posterior pole. Accumulation of lacZ mRNA continues to increase through gastrulation (D) and germband extension (E). Late stage, germband retracted embryos (F) show perdurance of expression in isolated cells, mostly located on the dorsal yolk sac. The embryos in this figure are from a mating of homozygous P{GAL4-nos.NGT}40 females with homozygous P{UAS-lacZ.B}4-1-2 males.
TABLE 1
Developmental time course of NGT-driven gene expression
Unique phenotypes produced by NGT-driven expression of different UAStransgenes: We examined the effects of using the strong P{GAL4-nos.NGT}40 line to drive expression of several different UAS transgenes in the blastoderm embryo. Embryonic lethality was observed for all but one of the different UAS lines tested. Cuticle preparations of the inviable embryos indicate that unique phenotypes are associated with NGT-driven expression of each of these genes. Ectopic expression of dpp produces embryos that lack ventral denticle belts (Figure 3A). This resembles the phenotypes obtained when dpp is ectopically expressed by other means in the early embryo and is consistent with the role that dpp has in dorsal-ventral patterning (Ferguson and Anderson 1992; Staehling-Hamptonet al. 1994). Patterning defects along the anterior-posterior axis are produced by NGT-driven expression of the segmentation genes en and runt. Uniform expression of the segment-polarity gene en in this manner generates a reproducible pair-rule phenotype (Figure 3B). Similar effects have been observed when en expression is induced by heat shock at the onset of gastrulation (Poole and Kornberg 1988). More severe patterning defects, including elimination of the entire head skeleton, are observed in embryos with NGT-driven runt expression (Figure 3C). The molecular basis of the phenotype produced by runt overexpression is discussed below.
We also obtained specific embryonic lethal phenotypes upon ectopic expression of lzand pnt, two genes that are most well characterized for their roles in pattern formation during eye development (Brunneret al. 1994; O’Neillet al. 1994; Dagaet al. 1996). NGT-driven expression of lz causes a dorsal closure defect similar to that observed in embryos mutant for the “tail-up” class of recessive lethal mutants (Figure 3D; Frank and Rushlow 1996). This tail-up phenotype seems unlikely to be relevant to a normal regulatory function, as lz shows only extremely limited expression during Drosophila embryogenesis (S. G. Kramer and J. P. Gergen, unpublished results). However, it is notable that this phenotype is distinct from that produced by expression of runt (compare Figure 3, C and D). These two genes encode members of the Runt domain family of transcription factors. The differential response of the embryo to these two related proteins provides a clear indication of their functional specificity, presumably due to differential interactions with other factors in the Drosophila embryo. Our experiments with pnt provide another example of a differential response, in this case to protein isoforms produced by differential splicing. Expression of the Pnt[P1] protein results in fully penetrant embryonic lethality with patterning defects in the head skeleton (Figure 3E). In contrast, the P{UAS-pnt.P2}2 transgene has no discernible effect on the viability or phenotype of embryos from homozygous P{GAL4-nos.NGT}40 females. The principal difference in these two transgenes is an N-terminal exon that confers constitutive activity on the Pnt[P1] protein, whereas Pnt[P2] is activated in response to receptor tyrosine kinase signaling pathways (Brunneret al. 1994; O’Neillet al. 1994). On the basis of these observations, we speculate that the defects caused by ectopic Pnt[P1] expression do not occur with Pnt[P2] because the level of receptor tyrosine kinase signaling activity within the cells that are affected by Pnt[P1] is insufficient to activate Pnt[P2].
Figure 3.
Embryonic lethal phenotypes associated with NGT-driven expression of different developmental regulators. Cuticle preparations of embryos produced in crosses between homozygous P{GAL4-nos.NGT}40 females and males carrying UAS-dpp (A), UAS-en (B), UAS-runt (C), UAS-lz (D), and UAS-pnt[P1] (E) transgenes. The embryos are oriented anterior end up.
TABLE 2
Effects of expressing various gene products with the maternal GAL4 system
Quantitative effects of NGT-driven expression: The above experiments are with P{GAL4-nos.NGT}40, one of the stronger maternal GAL4 drivers. We also examined the activities of other lines using similar assays. The results for three representative lines, P{GAL4-nos.NGT}11, P{GAL4-nos.NGT}31, and P{GAL4-nos. NGT}40, are shown in Table 2. Quantitation of β-galactosidase activity allows for a direct measurement of the relative levels of expression produced by these different lines. Even a single copy of P{GAL4-nos.NGT}11, the weakest line characterized in these experiments, drives expression of lacZ at levels that are 100-fold over background. The maternal effects are additive, as females that are homozygous for any particular NGT chromosome show approximately twice the level of activity as heterozygous females (Table 2). The relative strengths of these different NGT genotypes correlate with the lethality observed in crosses with different UAS transgenes (Table 2). Importantly, these data show that for a number of the toxic UAS transgenes tested in our assays, the difference between no discernible effect on viability and fully penetrant embryonic lethality occurs over a range that involves only a 4- to 5-fold increase in the level of NGT-driven expression.
We further characterized the basis of the lethality obtained in crosses with UAS-runt.One observation of particular interest was a pronounced sex bias in lethality that was not obtained with the other UAS transgenes. Males are consistently more sensitive than females to the lethal effects of NGT-driven runt expression. One explanation for this sex-biased lethality is that NGT-driven runt expression leads to inappropriate activation of the Sex-lethal (Sxl) gene in male embryos. This would be consistent with runt‘s role as a regulator of Sxl and would both confirm and extend recent results indicating that increased runt activity can result in transcriptional activation of Sxl in male embryos (Krameret al. 1999). However, we found no evidence that NGT-driven runt expression was capable of activating a Sxl[Pe]lacZ reporter gene in male embryos (data not shown). Furthermore, the preferential sensitivity of males to UAS-runt expression is not suppressed in males hemizygous for the Sxl mutations Sxl[F#1] and Sxl[7BO] (data not shown). If Sxl activation contributed to male lethality, then males carrying these loss-of-function mutations would show increased viability. Thus, we conclude thatSxl activation does not account for the enhanced sensitivity of males to UAS-runt expression.
An alternative explanation is that the activity of the P{UAS-runt.T} transgene is dosage compensated, i.e., it is twice as active in males as in females. Using the results of the β-galactosidase assays as a scale, we estimate that males are approximately twice as sensitive as females to the toxic effects of runt overexpression. For example, male viability is reduced to 10% in crosses with heterozygous P{GAL4-nos.NGT}31 mothers. A comparable reduction in female viability is obtained in crosses either with heterozygous P{GAL4-nos.NGT}40 mothers or with homozygous P{GAL4-nos.NGT}31 mothers. These two maternal genotypes are estimated to have twice the activity of the heterozygous P{GAL4-nos.NGT}31 mothers (Table 2). Similarly, fully penetrant lethality of the P{UAS-runt.T} transgene in males is obtained at half the level of NGT activity that is required for fully penetrant lethality in females (homozygous P{GAL4-nos.NGT}11 females vs.homozygous P{GAL4-nos.NGT}40 females, respectively). These results agree extremely well with a model whereby the embryonic activities of the P{UAS-runt.T} transgenes are dosage compensated. Previous work demonstrated that runt‘s activity during segmentation is dosage compensated (Gergen 1987). These results presented here strongly suggest that control elements responsible for dosage compensation of runt at this stage of development are contained within the P{UAS-runt.T} transgene.
Quantitative response of target genes to UAS-runt during segmentation: The expression of genes in both the pair-rule and segment-polarity classes of segmentation genes is altered in runt mutant embryos (Carroll and Scott 1986; DiNardo and O’Farrell1987; Ingham and Gergen 1988; Baumgartner and Noll 1990). The identification of the genes that are direct targets for transcriptional regulation by the Runt protein is complicated by a number of cross-regulatory interactions between these different genes. One approach that has been used extensively to investigate the regulatory circuitry in the segmentation pathway has involved ectopic expression using heat-inducible transgenes. For example, runt‘s roles in activating the pair-rule gene ftz and repressing the pair-rule genes eve and hairy have been investigated using hs-runttransgenes (Tsai and Gergen 1994, 1995). Additional experiments with these hs-runt lines have also suggested that runt directly regulates segment-polarity gene expression (Manoukian and Krause 1993).
We examined the response of several segmentation genes to different levels of NGT-driven runt expression. To obtain insight on the basis for the lethality associated with NGT-driven runt expression, we examined embryos from a cross of homozygous P{GAL4-nos.NGT}11 females with homozygous P{UAS-runt.T} males. The level of UAS-runtactivity in embryos from this cross is fully lethal to males and allows for only limited female viability (Table 2). Embryos from this cross show a consistent repression of the odd-numbered stripes of the segment-polarity gene en (Figure 4F). In contrast, expression of the pair-rule genes eve and ftz is not altered in these same embryos (Figure 4, G and H). Similarly, this level of ectopic runt activity has little to no effect on the expression of the pair-rule genes hairy, paired, odd-skipped, and sloppy-paired (data not shown). This strongly suggests that the repression of en is not mediated indirectly through alterations in the expression of these other pair-rule genes, and it provides evidence that runt directly represses the odd-numbered en stripes. As expected, some variation is also observed in the patterns of en expression. All embryos show defects in the initiation of the odd-numbered stripes during the early stages of germband extension. In later stage embryos, the expression of the odd-numbered en stripes is partially or even fully restored. Approximately 10% of the embryos at full germband extension show apparently normal en expression. This corresponds well to the proportion of progeny from this cross that will survive to adulthood.
We also examined segmentation gene expression patterns in embryos from a cross of homozygous P{GAL4-nos.NGT}40 females with homozygous P{UAS-runt.T} U15 males. This combination is estimated to allow for an approximately sixfold higher level of ectopic expression than that obtained in the above experiment and is equal to or greater than that obtained within the pair-rule stripes produced by the endogenous runt gene (Figure 4I). As observed at lower levels, the odd-numbered stripes of en are repressed by this higher level of ectopic runt expression (Figure 4J). The effects of high-level NGT-driven runt expression on the pair-rule genes mimic what has been described previously in hs-runt embryos. Expression of eve is reduced and there is a difference in the sensitivity of the different stripes, stripe 2 being the most sensitive to repression by runt (Figure 4K). Similar stripe-specific repression of hairy stripe #1 is also observed in these embryos (data not shown). Expression of ftz is increased with the broader stripes fusing, especially in the more posterior regions (Figure 4L). These findings validate previous results with hs-runt embryos and indicate that these different regulatory interactions are not a result of the physiological perturbations associated with the heat-shock response. The localized effects of NGT-driven runt expression on the pair-rule genes contrast with the relatively uniform repression of the odd-numbered en stripes. This provides a further indication that the repression of en is direct and not mediated by other pair-rule genes.
DISCUSSION
A maternal GAL4 system: We have developed and characterized Drosophila strains that express the yeast transcription factor GAL4 maternally. These strains fill a void in the collection of GAL4 drivers that are available for manipulating gene expression in Drosophila. Many of the GAL4 lines that have been characterized previously are based on the initial expression constructs of Brand and Perrimon (1993). A key difference in the GAL4 expression construct described here is the use of the 3′ UTR region of the αTub84BmRNA in place of the hsp70 terminator. The use of the tubulin 3′ UTR also distinguishes the NGT strains from GAL4 drivers that use the 3′ UTRs of either bicoid (Arnostiet al. 1996) or nos (Van Dorenet al. 1998) to deliver maternally expressed transcripts specifically to the anterior or posterior pole of the embryo, respectively. The spatially uniform activation mediated by the NGT drivers simplifies the quantitative interpretation of experimental results both in the entire embryo as well as on the cellular level.
Figure 4.
Effects of maternal GAL4-driven runt expression on segmentation gene expression. The top row of embryos shows the normal mRNA expression patterns of runt (A), en (B), eve (C), and ftz (D) as revealed by whole-mount in situhybridization. The alterations in expression of these same genes in embryos from a cross of homozygous P{GAL4-nos.NGT}11 females with homozygous P{UAS-runt.T}232 males are shown in E, F, G, and H, respectively. Similarly, I, J, K, and L show the respective expression of these four genes in embryos from a cross of homozygousn P{GAL4-nos.NGT}40females with homozygous P{UAS-runt.T}U15 males. Embryos are oriented anterior to the left, dorsal side up. All embryos are in the blastoderm stage, except for those probed for en expression (B, F, and J), which have completed cellularization and have gastrulated. The circles in F and J indicate regions where the odd-numbered en stripes are absent. Similarly, the circle in K marks the absence of eve stripe 2.
The importance of mRNA control elements in GAL4 misexpression strategies is also reemphasized by the recent work of Rørth (1998). In this case it was found that GAL4-driven expression during oogenesis was qualitatively altered by including the 3′ untranslated region of the maternally expressed K10 gene in place of the SV40 polyadenylation signal in the UAS expression construct (Rørth 1998). It seems likely that further modifications of the 5′ and 3′ UTRs of both the GAL4 driver and UAS responder will allow for even greater control over gene expression during oogenesis and early embryogenesis. It may be possible to increase blastoderm stage expression by including the 3′ UTR from a mRNA that is efficiently translated during this stage in the UAS expression construct. Similarly, it may also be possible to greatly reduce the postblastoderm expression by including elements from mRNAs that are turned over rapidly during these stages in both the NGT– and UAS expression constructs. Incorporating the results of further studies on mRNA control elements should greatly improve the specificity of this type of strategy for manipulating gene expression.
There are several attractive features of using maternally driven GAL4 expression to investigate gene function. First, the large body of information on the genetics and cell biology of the Drosophila embryo makes this a powerful developmental context for interpreting the phenotypes produced by ectopic expression of any target gene of interest. Experiments at this stage of development also avoid some of the difficulties that can arise with other GAL4 drivers that have low levels of expression at stages or in tissues other than the desired developmental context. With a maternal GAL4 driver, the earliest stage at which a paternally inherited UAS transgene is available for activation by GAL4 protein is within the fertilized egg. Finally, the relative simplicity of the ectopic expression pattern makes it relatively straightforward to interpret the experimental results quantitatively. The three NGT lines characterized here allow for control of expression over a range that differs >10-fold between the weakest line as a heterozygote and the strongest line as a homozygote (Table 2). Using enzyme activity measurements as a benchmark, we estimate this range to extend from ~10,000 to 125,000 molecules of β-galactosidase/cell at the blastoderm stage. Table 2 does not include data from other, weaker lines. It is also possible to obtain higher expression levels by combining different NGT drivers. Altogether, with the NGT lines that are currently available, we estimate being able to manipulate expression levels over a range that approaches two orders of magnitude.
The expression levels obtained with the different NGT drivers are presumed to be due to the sites of transgene insertion. Similar position effects are, of course, also observed for different inserts of any given UAS transgene. Indeed, the ability to mix and match NGT drivers with different UAS responders provides an additional level of flexibility that can be useful in the design and interpretation of ectopic expression experiments. Our experiences with several different P{UAS-runt.T} lines indicate a robust and linear response over the ranges tested. The phenotype observed in any given cross depends on the level of runt expression that is obtained, irrespective of the particular combination of NGT driver and UAS-runt responder that is used in the cross.
There is an additional point to be made with respect to our attempts to quantitatively characterize this ectopic expression assay system. As indicated in the tables, the standard error in the measurements of NGT-driven β-galactosidase activity averages ~20% of the activity measured, irrespective of the total absolute enzyme activity. These assays were carried out on individual embryos that were staged based on their time of gastrulation, a readily observed and rapid morphogenetic change that immediately follows the completion of cellularization. Thus, the embryo-to-embryo variability is not likely to be due to differences in developmental stage. Repeated measurements on extracts from different single embryos further indicate that differences in micropipeting do not account for this level of variability. Thus, the variability may be intrinsic within this biological system. This variation somewhat complicates the utility of this system for confident measurement of small (<20%) changes in the level of gene function. However, the ability to manipulate gene expression in a stepwise manner over one to two orders of magnitude with a reliability that approaches ±20% will provide an important tool for quantitative analysis of gene function in vivo. Indeed, the importance of quantitative considerations is emphasized by the dose-dependent effects obtained with several different toxic UAS transgenes. In each case we found a relatively sharp threshold in the biological response, with the difference between no apparent phenotype and a nearly fully penetrant, lethal phenotype resulting from a four- to fivefold increase in the level of NGT-driven ectopic gene expression.
Interactions between runt and the sex determination pathway: Our results reveal that male embryos are reproducibly more sensitive than females to the toxic effects of UAS-runt expression. Previous work reveals there are multiple interactions between runtand the sex determination system in Drosophila. The master regulator in the sex determination pathway is the Sxl gene, which normally is activated in females and repressed in males. runt plays a role in the transcriptional activation of Sxl, and increasing the dosage of runt activity is sufficient for triggering the inappropriate activation of Sxl in preblastoderm male embryos (Krameret al. 1999). Inappropriate activation of Sxl could in principle account for the preferential male lethality observed with NGT-driven runt expression. However, male lethality is not suppressed in males that are mutant for Sxl. Furthermore, NGT-driven runt expression is not capable of activating the full-length SxlPe embryonic promoter in males. Presumably the levels of ectopic runt expression obtained during the syncytial blastoderm stage are not high enough to trigger inappropriate activation of the Sxl embryonic promoter in males. It remains to be determined whether the levels of NGT-driven UAS-runt expression are sufficient for SxlPe activation in females.
The preferential sensitivity of males to NGT-driven runt expression is readily explained if P{UAS-runt.T} activity is dosage compensated. A comparison of the NGT expression levels required to produce comparable effects on male and female viability is consistent with a twofold increase in UAS-runt potency in males. These observations provide the basis for a model in which Sxl, which is specifically expressed in females, buffers the embryos from the toxic effects of UAS-runt expression. In addition to its more widely known role in regulating splicing, the Sxl protein represses expression of msl-2 by interacting directly with the 5′ and 3′ UTR of the msl-2 mRNA (Bashaw and Baker 1997; Gebaueret al. 1998). The runt 3′ UTR contains several putative Sxl binding sites (Kelleyet al.1995) which may mediate translational repression by the Sxl protein, and this entire region is contained within the P{UAS-runt.T} transgenes used in our experiments. In support of this hypothesis, UAS-runt transgenes that use the SV40 3′ UTR instead of the runt 3′ UTR show similar, though reduced, levels of activity in both sexes (J. Wheeler, D. Swantek and J. P. Gergen, unpublished results). Further work is needed to confirm the mechanisms responsible for the enhanced potency of UAS-runt transgenes in males. Indeed, the activity of UAS-runt transgenes may provide a useful tool for further investigation of this mode of dosage compensation.
Dose-dependent effects of runt on segmentation: The Runt protein is a pivotal transcriptional regulator in the pathway of segmentation in Drosophila. Previous work with heat-inducible hs-runt transgenes has indicated that Runt functions to activate the transcription of some downstream targets, such as ftz, while repressing the transcription of targets such as en, eve, and hairy (Tsai and Gergen 1994, 1995). Furthermore, there are at least two mechanisms for transcriptional repression by the Runt protein. The stripe-specific repression of eve and hairy involves interactions with the corepressor protein Groucho that are mediated by a conserved VWRPY motif located at the C terminus of the Runt protein (Aronsonet al. 1997). In contrast, the repression of the segment polarity gene en and the head gap gene orthodenticle occurs through a VWRPY-independent mechanism (Aronsonet al. 1997; Tsaiet al. 1998). The results obtained here using maternally provided GAL4 to drive ectopic runt expression confirm these previous findings and indicate that these varied regulatory effects are not an artifact due to overexpression by heat-shock treatment. An important additional finding is that the odd-numbered stripes of the segment-polarity gene en are efficiently repressed in embryos that display normal pair-rule gene expression patterns. This is strong evidence that the repression of these en stripes is not indirectly mediated through Runt’s regulatory effects on the expression of these other pair-rule genes. These results strongly suggest that the odd-numbered en stripes are a direct target for repression by the Runt protein.
What is the relevance of runt‘s ability to repress the odd-numbered en stripes for normal segmentation? The even- and odd-numbered en stripes are controlled by two distinct regulatory programs (DiNardo and O’Farrell 1987; DiNardoet al. 1988). The even-numbered stripes emerge first and form in the center of the seven stripes of runtexpression that are present during the blastoderm stage. The odd-numbered en stripes emerge during the process of cellularization and are not expressed at levels comparable to the even-numbered stripes until after gastrulation (DiNardo and O’Farrell1987). The odd-numbered en stripes arise in the regions between the runt stripes and in fact appear to define the center of the runt interstripes. The runt stripe/interstripe pattern develops from an earlier broad band of expression throughout the presegmental region of the embryo (Klingler and Gergen 1993). The levels of NGT-driven runt expression in the presumptive interstripe regions that are required for repression of the odd-numbered en stripes are below the levels of expression within the runtstripes (Figure 4, E and F) and are comparable to the levels obtained during the earlier broad band stage. These observations suggest that clearance of Runt from the interstripes is an essential prerequisite for the formation of the odd-numbered enstripes.
Among the segmentation genes examined here, en is the most sensitive to ectopic runtexpression. Indeed, the threshold response of en correlates extremely well with the threshold for the lethality that is associated with NGT-driven runt expression. This lethal phenotype provides a powerful entry point for a genetic dissection of Runt function. One approach is to screen for mutations that suppress the lethality associated with runtoverexpression during this early stage of embryogenesis. There is an obvious practical advantage of genetic screens that are based on suppression of a lethal phenotype. There is also an important theoretical advantage of this kind of approach compared to sensitized screens that are based on the enhancement of intermediate mutant phenotypes. Mutations with generally deleterious effects would be expected to increase the severity of defects in a sensitized genetic screen based on enhancement of loss-of-function phenotypes, but in a nonspecific manner. However, these same mutations also would be expected to enhance, not suppress, the severity of defects produced by GAL4-driven overexpression. From our quantitative analysis, we feel it should be possible to identify and characterize mutations that have a twofold effect on the potency of the Runt protein. In summary, the ability to quantitatively manipulate gene expression in the Drosophila blastoderm embryo should provide a powerful new tool for genetic studies on the function of runt as well as any other regulatory gene product that can function within this well-defined developmental context.
Acknowledgments
The excellent and invaluable technical assistance of Claudia Brunner and Deborah Swantek is greatly appreciated. Joe McLean provided the impetus and some assistance with cuticle preparations of embryos carrying various UAS transgenes. This manuscript was improved by comments from John Wheeler, Kathy Wojtas, and Christine Vander-Zwan. Recombinant DNA plasmids containing the nos promoter and the αTub84B 3′ UTR were generous gifts from Liz Gavis (Princeton, NJ) and Bill Theurkauf (University of Massachusetts), respectively. The starting vectors for the GAL4 modular misexpression system, pGAT-B and pUAS-T, were provided by Andrea Brand (Welcome MRC Institute, Cambridge, United Kingdom). The P{UAS-lz} flies were a gift from Utpal Banerjee (UCLA). Sxl mutations were obtained from Tom Cline (Berkeley) and Jim Ericson (Columbia). Many of the other Drosophila lines used in this work were obtained from the stock center in Bloomington. This work was supported by a National Institutes of Health grant GM-53229 to J.P.G.
Footnotes
