raw mediates antagonism of ap-1 activity in drosophila mediates antagonism of ap-1 activity in...

Copyright � 2008 by the Genetics Society of AmericaDOI: 10.1534/genetics.107.086298

Raw Mediates Antagonism of AP-1 Activity in Drosophila

Katherine L. Bates,1 Matthew Higley and Anthea Letsou2

Department of Human Genetics, University of Utah, Salt Lake City, Utah 84112

Manuscript received December 25, 2007Accepted for publication February 3, 2008

ABSTRACT

High baselines of transcription factor activities represent fundamental obstacles to regulated signaling.Here we show that in Drosophila, quenching of basal activator protein 1 (AP-1) transcription factoractivity serves as a prerequisite to its tight spatial and temporal control by the JNK ( Jun N-terminal kinase)signaling cascade. Our studies indicate that the novel raw gene product is required to limit AP-1 activity toleading edge epidermal cells during embryonic dorsal closure. In addition, we provide the first evidencethat the epidermis has a Basket JNK-independent capacity to activate AP-1 targets and that raw function isrequired broadly throughout the epidermis to antagonize this activity. Finally, our mechanistic studies ofthe three dorsal-open group genes ½raw, ribbon (rib), and puckered (puc)� indicate that these gene productsprovide at least two tiers of JNK/AP-1 regulation. In addition to Puckered phosphatase function in leadingedge epidermal cells as a negative-feedback regulator of JNK signaling, the three dorsal-open group geneproducts (Raw, Ribbon, and Puckered) are required more broadly in the dorsolateral epidermis toquench a basal, signaling-independent activity of the AP-1 transcription factor.

ACTIVITIES of the three signaling subfamilies ofthe MAP kinase (MAPK) signaling pathway: ex-

tracellular signal-related kinase (ERK), p38, and JunN-terminal kinase ( JNK), are tightly controlled bothspatially and temporally in numerous experimental sys-tems and organisms (Qi and Elion 2005). It is well-documented that threshold MAPK signaling activities aremaintained by a balance of activators (MAPK kinases)and negative-feedback regulators (MAPK phosphatases).Less well considered, however, are the effects of base-line MAPK signaling levels and/or pulses of signalingmolecules that are produced in stochastic fashion(‘‘signaling noise’’) (Rao et al. 2002).

Both signaling baselines and signaling noise repre-sent theoretical obstacles to attaining tightly controlledand/or robust responses to genetically programmedactivation of signal cascades such as the MAPK pathway.First, a baseline of signaling activity close to a theoreticalactivation threshold requires a responding cell to dif-ferentially recognize and interpret comparatively smallchanges in signaling activities. Second, signaling noise,which occurs when various gene products fluctuateabout their average values, might drive a process tothreshold activation by mere chance. The probability ofnoise-driven activation is greatest when a baseline ofsignaling activity is already close to a theoretical activa-tion threshold. Indeed, signaling noise is responsible

for many well-characterized stochastic biological events,including, for example, stem cell division and retrovirallatency (Hasty et al. 2000; Lemischka 2005). For non-stochastic signaling outcomes, however, theoretical stud-ies dictate genetic programs which ensure minimalsignaling baselines and buffered noise. Experimentalmethods to approach this problem in vivo have beenproblematic as the gain-of-function phenotypic con-sequences of ectopic signaling are not always easilypredicted.

To genetically dissect the control of MAPK baselinesignaling activities, we turned to the Drosophila JNKsignaling cascade, a pathway for which the phenotypicconsequences of ectopic signaling are understood. Inconjunction with its associated transcription factor Jun/Fos (activator protein 1, AP-1), the JNK cascade functionsin the fruit fly Drosophila melanogaster as the essentialMAPK regulator of several developmental and physio-logical processes. These include tissue morphogenesis,wound healing, planar cell polarity, programmed celldeath, and synaptic plasticity, as well as the immune andoxidative stress responses (for review see Stronach

2005). Among these processes, it is during tissue mor-phogenesis, and embryonic dorsal closure in particular,that the role for JNK signaling and the mechanisms ofits regulation have been best characterized.

In Drosophila, embryos undergo dorsal closure be-tween 8 and 12 hr after egg lay (AEL). Closure com-prises three phases: (1) initiation, (2) spreading, and(3) suturing. At the initiation of closure, leading edge(LE) epidermal cells elongate along their dorsoventralaxis. During the spreading (or sweeping) stage of dorsalclosure, epidermal cells positioned behind the LE also

1Present address: California State University, Dominguez Hills, Carson,California 90747.

2Corresponding author: Department of Human Genetics, University ofUtah 15 N. 2030 E., Room 2100, Salt Lake City, UT 84112.E-mail: [email protected]

Genetics 178: 1989–2002 (April 2008)

elongate along their dorsoventral axis, and thus theepidermis spreads dorsally to cover the embryonicdorsal surface. In the final stage—suturing—the epi-dermal sheets approach one another at the embryonicdorsal midline. Closure is complete when epidermalsheets suture at the dorsal midline. As it is the epidermisthat secretes the larval cuticle, mutations in loci re-quired for closure (dorsal-open group loci) are recog-nized by their embryonic lethal cuticular phenotype, asingle large dorsal hole.

Molecular analyses of the dorsal-open group loci haverevealed that these code for three classes of gene prod-uct. The class I and II dorsal-open group genes, respec-tively, code for JNK/AP-1 and Dpp signaling components;the class III genes code for structural molecules.

Accumulated data from several labs have led to ourunderstanding that the JNK and Dpp signaling pathwaysfunction sequentially to coordinate epidermal spread-ing during embryonic dorsal closure (for review seeKnust 1997). First, an as yet uncharacterized triggeractivates JNK/AP-1 signaling in LE epidermal cells.Next, JNK/AP-1 signaling induces expression of targetgenes, including dpp and puc, in LE cells. Whereas thepuc-encoded phosphatase feeds back to negatively reg-ulate JNK/AP-1 signaling in a cell-autonomous fashion,the dpp-encoded cytokine function is non-cell autono-mous. Upon secretion from LE cells, Dpp activatessignaling broadly in epidermal cells and thereby indu-ces their elongation. Regulation of the JNK/AP-1 andDpp signaling pathways is tightly integrated in dorsalclosure, but only the JNK/AP-1 pathway is autoregula-tory. Although dpp can autoactivate its own expression atsome embryonic sites ( Jazwinska et al. 1999b), it doesnot do so in LE epidermal cells (Arora et al. 1995).

At the cellular level, JNK/AP-1 signaling leads to re-organization of cytoskeletal proteins in LE cells of theepidermal sheets. Whereas actin-, myosin-, and phos-photyrosine-containing proteins accumulate at the LEin wild-type embryos, their accumulation is disrupted inembryos deficient in the AP-1 components Jun ( Jun-related antigen/Jra), or Fos (Kayak/Kay), or in the con-stituents of the relay system that function to potentiateAP-1 activity via phosphorylation ½ JNKKK (Misshapen/Msn), JNKKK (Slipper/Slpr), JNKK (Hemipterous/Hep), JNK (Basket/Bsk)�. That mutations in the genescoding for each of these signaling activators lead to anidentical dorsal-open phenotype highlights the non-redundant contribution of each to dorsal closure (forreview see Stronach 2005).

A somewhat surprising discovery, but one that none-theless underscored the requirement for precise con-trol of JNK/AP-1 signaling during dorsal closure, wasthe finding that mutations in JNK/AP-1 activators andJNK/AP-1 antagonists both give rise to dorsal-open em-bryonic lethal phenotypes. As discussed in the currentreport and elsewhere (Byars et al. 1999; Bradley andAndrew 2001), mutations in three genetically defined

dorsal-open group loci (raw, ribbon, and puckered) leadnot to the absence of LE JNK/AP-1 signaling, but ratherto its ectopic activity in several rows of the dorsolateralepidermis.

The best characterized of the dorsal-open groupsubset of JNK/AP-1 antagonists is puckered (puc), whichencodes a VH1-like dual-specificity protein tyrosinephosphatase belonging to the MAPK subfamily ofmitogen kinase phosphatases (MKPs) (Martin-Blanco

et al. 1998). MKPs are characterized by highly conservedprotein domains that include the protein tyrosinephosphorylation (PTP) loop and two N-terminal CH2domains (Keyse and Ginsburg 1993; Muda et al. 1996;Tanoue et al. 2000). It is thought, on the basis of thesesequence homologies, that the Drosophila MKP Puck-ered acts by dephosphorylating and thereby inactivatingthe basket-encoded AP-1 activator JNK (Martin-Blanco

et al. 1998). More specifically, puc is expressed in LE cellsduring dorsal closure and is thought to be part of afeedback loop controlling the magnitude of epidermalJNK signaling. There is not currently a comprehensiveexplanation for why loss of an intracellular JNK antag-onist leads to ectopic AP-1 activity, although it is formallypossible that an unidentified secreted JNK-signalingtarget feeds forward to activate the JNK pathway inparacrine fashion (Knust 1997). For the studies re-ported here, we considered an alternative mechanismfor ectopic signaling in raw (and puc and rib) mutantembryos—that a second regulatory network directlyrepresses basal JNK and/or AP-1 activities in the em-bryonic dorsolateral epidermis.

As we report here, our studies of the Drosophiladorsal-open group raw gene reveal an unexpectedBasket JNK-independent AP-1 activity in the embryonicdorsolateral epidermis at the time of dorsal closure andidentify an essential role for raw in regulating thisactivity. Our findings provide a compelling explanationfor dysregulated signaling in raw (and puc and rib)mutant embryos and have important implications forunderstanding mechanisms of AP-1 regulation duringdevelopment in flies and higher eukaryotes. In partic-ular, our studies indicate that the role of raw is primarilyto eliminate Basket JNK-independent activity of the AP-1 transcription factor in the embryonic epidermis andthereby establish an AP-1-silenced ground state prior tothe onset of dorsal closure and secondarily, to restrictactivity of the JNK transcriptional effector AP-1 to the LEsubset of epidermal cells during dorsal closure. Further-more, our studies indicate that raw functions as ageneral antagonist of AP-1 activity, required for at leastthree developmental occurrences of JNK/AP-1 signal-ing in Drosophila. Taken together, our data lead us tosuggest that the biological effect of raw-mediated AP-1antagonism is to establish a sensitized state of compe-tence; this condition then sets the stage for a highlyrobust, Basket JNK-dependent activation of AP-1 tran-scriptional activity.

1990 K. L. Bates, M. Higley and A. Letsou

MATERIALS AND METHODS

Fly stocks: raw1, raw2, raw2418, rawlex1, rawlex2, bsk1, bsk2, JraIA109,pucE69, pucH246, rib1, and w1118 mutant lines, balancer lines, andUAS-dpp and 69B-gal4 transgenic lines have been described(Byars et al. 1999; FlyBase Consortium 2003). We obtainedUAS-brk transgenics from B. Stronach, Kr-gal4 transgenics fromM. Lamka, LE-gal4 transgenics from S. Noselli, and GMR-gal4transgenics from K. Broadie. Double-mutant lines weregenerated using standard mating strategies and recombina-tion when necessary. All our lines were tested to ensure thatthe desired mutations were present. To do this, we assayed forlethality in trans to independently derived alleles of each of themutants used in our study. All fly lines were maintained onstandard cornmeal/molasses/agar medium at 22�, unlessotherwise stated.

Phenotypic analyses: Embryonic lethal cuticular pheno-types were viewed after mounting devitellinized samples inone-step mounting medium ½30% CMCP-10 (Masters Chem-ical Company), 13% lactic acid, 57% glacial acetic acid�.Hybridizations in situ using digoxigenin-labeled RNA probeswere performed as described (Byars et al. 1999). For all ourphenotypic studies, we looked at no fewer than 50 mutantsamples after development at constant temperature. Embryosand cuticles were analyzed and imaged using dark field, brightfield, and DIC optics. In all cases, included figure images arerepresentative of a large experimental set.

Epistasis studies: Mutants included in our epistasis studieshave been defined either molecularly or genetically as null. Asgenetic definitions of allele strength are sometimes contro-versial, we note here that it is sometimes appropriate to usehypomorphs in double-mutant epistasis studies (Huang andSternberg 1995). In particular, for epistasis studies of twogenes A and B, a gene A hypomorph may be used when B isepistatic to A. In this scenario, the normal function of A wouldbe to negatively regulate B, and AB double-mutant animalswould not have functional B activity that could be regulatedby residual A activity. We also note that it is appropriate to useeither an amorph or a severe loss-of-function hypomorphwhen a gene has an earlier maternal (and/or zygotic) re-quirement in development, as long as the chosen allelecompletely disrupts the process being studied (Huang andSternberg 1995).

Rescue of raw mutant embryonic lethality: To assess rescueof raw lethality in a tissue-specific manner, lethal phaseanalyses of embryos derived from matings of w1118; raw2418/CyO, Ub-GFP; p[UAS-raw1] females to w1118; raw2418/CyO, Ub-GFP; gal4 or w1118; raw2418 gal4/CyO Ub-GFP males were per-formed. We employed four independent GAL4 lines to driveexpression of the UAS-raw1 transgene in raw2418 homozygotes.Rescued raw2418 homozygous first instar larvae were identifiedby their inability to fluoresce. raw2418/raw2418 viability was con-firmed by single larva PCR using primers directed against GFP.

To assay rescue of raw-dependent lethality, 4-hr collectionsof embryos derived from matings of w1118; raw2418/CyO, Ub-GFP;p[hs-raw1] flies were aged as indicated in the text and thensubjected to a 1-hr 37o heat shock. After 36 hr, newly hatchedlarvae were examined for the presence or absence of the GFP-marked balancer chromosome using fluorescent optics.raw2418 homozygotes were identified as above.

Plasmid constructions: P-element rescue constructs pCaSper-raw and pUAST-raw were generated from a full-length (MfeINotI ) raw cDNA (Byars et al. 1999). FLAG-tagged raw isoformswere generated by insertion of a 59FLAG sequence for pUAST-rawFLAG, and by insertion of the MfeI NotI fragment of raw ina pFLAG-CMV-5b (Sigma, St. Louis) derivative in which theApaLI site was converted to KpnI using the linker TGCATCTCGGTACCC (pFLAG-CMV-raw).

Cell transfection and immunofluorescence: Anti-FLAG M2(1:100, Sigma) and anti-c-Myc 9E10 (1:250, Santa Cruz Bio-technology) antibodies were used for immunostains in whole-mount embryos and cell culture. Secondary antibodies anddyes used for this study included anti-mouse FITC (1:200,Jackson Labs), anti-mouse Cy3 (1:500, Jackson Labs), andDAPI ( Jackson Laboratories, West Grove, PA). COS-7 cells,grown in a monolayer at 37� in 5% CO2, were transientlytransfected using the standard calcium method and fixed byacetone/methanol (Debry et al. 1997).

RESULTS

raw mutants exhibit ventral cuticular patterning de-fects that are attributable to ectopic signaling by Dpp:Our previous phenotypic raw study drew our attentionto two highly atypical features (with respect to thedorsal-open group) of the raw mutant phenotype(Byars et al. 1999). First, we found that in addition toarchetypal cuticular defects marking abnormalities indorsal closure, cuticles derived from raw mutant em-bryos exhibit ventral cuticular patterning defects, andsecond, we observed ectopic JNK/AP-1 signaling in thelateral epidermis of dorsal closure stage raw mutantembryos. Deciphering this pleiotropy has been centralto our current understanding of Raw function and AP-1antagonism.

As an initial step toward a better understanding ofRaw, we defined an allelic series of mutations that wasbased upon the severity of the dorsal-closure defect andthen we assessed the associated raw-dependent pheno-types. The severity of raw-dependent ventral cuticularpatterning defects is positively correlated with the de-fining defect in dorsal closure. In cuticles derived fromembryos homozygous for the null raw1 allele, ventraldenticle belts are nonexistent (Figure 1A). As dorsalclosure defects are diminished, ventral denticle beltdysmorphogenesis is similarly diminished; in cuticlesderived from embryos homozygous for the hypomor-phic raw2 and raw2418 alleles, ventral denticles are evi-dent, however, while properly positioned and shaped,denticles are smaller and less refractile to light, and thebelts themselves do not extend as far dorsally in com-parison to wild type (Figure 1, C and E). In homozygotesof the weakest allele of raw, rawlex2, ventral denticles arewild type in position, shape, and form (Figure 1G; seealso Figure 3B).

Next, we assessed the extent of ectopic JNK/AP-1signaling in raw mutant embryos. In this regard, ourdemonstration that dpp and puc, two transcriptionallyregulated targets of JNK/AP-1 signaling during closure,are ectopically expressed in raw1 mutant embryos wasintegral to our characterization of Raw as a JNK/AP-1signaling antagonist. Whereas dpp and puc expressionin wild-type embryos is limited to a single row of cells,corresponding to the LE of the epidermis, dpp and pucexpression in the null mutant raw1 extends from the LEapproximately nine cell widths into the lateral epider-

raw Antagonism of AP-1 1991

mis (compare Figures 1B and 3A; see also Byars et al.1999).

In an extension of these studies, here we used dppexpression to monitor epidermal JNK/AP-1 activity inseveral different raw mutants. We found that as theventral denticle phenotype is correlated with cuticularhole size, so also is the extent of ectopic JNK/AP-1signaling in the dorsal epidermis. dpp expansion in theepidermis of raw1 mutant embryos represents the great-est degree of expansion for any of the raw alleles (Figure

1B). In embryos homozygous for raw2 and raw2418, allelesdefined via dorsal and ventral cuticular defects to be ofintermediate strength, dpp expression expands laterallyonly four to six cells (Figure 1, D and F). In the weakestallele of raw, rawlex2, dpp expansion is minimal, extendingat most to one additional row of epidermal cells (Figure1H). Thus, on the basis of the ectopic expression of dppas a readout for epidermal JNK/AP-1 activity, we con-clude that the strength of the raw phenotype correlateswith the extent of expanded signaling activity. Notably,it is the dorsal-most epidermis, not just the LE, whichis competent for JNK- and/or AP-1-mediated geneexpression.

Studies reported by us and others revealed the ca-pacity of pan-epidermal Dpp to disrupt ventral cuticularpattern (Staehling-Hampton and Hoffmann 1994;Riesgo-Escovar and Hafen 1997a; Byars et al. 1999).In particular, we have shown that although closureoccurs normally in UAS-dpp/69B-gal4 transgenics ex-pressing dpp ectopically throughout the embryonicepidermis, ventral cuticular differentiation in theseanimals is abnormal. Transgenics suffer an embryoniclethality that is associated with the same poorly dif-ferentiated ventral cuticle that distinguishes the rawmutant from the dorsal-open group more generally.Notably, the UAS-dpp/69B-gal4-associated cuticle is notdorsalized as (1) dorsal hairs (cuticular markers ofdorsal fate) do not extend to ventral cuticular domains,and (2) gastrulation (monitored visually in living em-bryos) proceeds normally (Byars et al. 1999).

In a complementary test of dpp effects in the dorsalepidermis of raw mutant embryos, here we employedthe UAS-Gal4 system to express the dpp antagonistbrinker (brk) pan-epidermally. Our phenotypic charac-terization of cuticles derived from raw1/raw1; UAS-brk/69B-gal4 mutants revealed rescue of raw1-dependentventral cuticular defects, but no rescue of the associateddorsal-closure defects (Figure 2, A–C). Comparison ofbrk expression in wild-type and raw mutant embryosfurther exposed the extent to which the epidermis ismispatterned in raw mutant embryos. In wild-typeembryos undergoing dorsal closure brk is expressed ina wide stripe of lateral epidermis just ventral to the LE,as well as in a more ventrally positioned epidermal stripe(Figure 2D). In contrast, in raw1 homozygotes both thelateral and the ventral epidermal stripes of brk expres-sion are missing while internal (nonepidermal) do-mains of expression are unaffected (Figure 2E). Takentogether, these data demonstrate that it is Dpp and/orBrinker (but not the JNK/AP-1 signaling cascade) thatis responsible for the ventral cuticular defects observedin raw mutant embryos.

A subset of dorsal-open group genes (the raw group)exhibits ventral cuticular patterning defects: Havingdefined a palette of raw-dependent embryonic pheno-types, we recognized that two other dorsal-open groupmutants (puc and rib) exhibit dpp expression patterns

Figure 1.—raw mutants define an allelic series. (A, C, E,and G) Embryonic cuticles and (B, D, F, and H) dpp mRNAexpression in stage 13 whole-mount embryos. (A) Cuticles de-rived from embryos homozygous for raw1, a null allele, exhibitsignificant dorsal closure and head involution defects, as wellas an absence of ventral denticle belts (*); (B) in whole-mountembryos in situ, leading edge dpp expression extends to adepth of nine cells in the lateral epidermis (Y). (C) Cuticlesderived from embryos homozygous for the raw2 allele, astrong hypomorph, also exhibit significant dorsal closureand head involution defects, as well as severely atrophied ven-tral denticle belts (*). (D) In whole-mount embryos in situ,dpp expression extends to a depth of seven cells in the lateralepidermis (Y). (E) Cuticles derived from embryos homozy-gous for the raw2418 allele, a hypomorph of intermediatestrength, exhibit a strong dorsal pucker, in addition to hyper-trophied ventral denticle belts (*). (F) In whole-mount em-bryos in situ, dpp expression extends to a depth of fourcells in the lateral epidermis (Y). (G) Cuticles derived fromembryos homozygous for rawlex1, a very weak hypomorph, ex-hibit a weak dorsal pucker and ventral denticle belts that areindistinguishable from wild type (*). (H) In whole-mount em-bryos in situ, dpp expression only minimally extends into thelateral epidermis, at most to a depth of two cells laterally (Y).In this and all subsequent figures, embryos are oriented suchthat dorsal is up and anterior is to the left.


and ventral cuticular defects analogous to those that wehad observed in raw mutants. Albeit variably expressed,the shared loss-of-function phenotypes are notable, andwe reasoned that insights into the function of the novelRaw protein in regulating JNK/AP-1 signaling might begarnered from a molecular and genetic understandingof puc and rib, by themselves and in conjunction withraw. puc codes for a MAPK phosphatase (Martin-Blanco et al. 1998), and rib codes for a putative BTB-POZ-type DNA binding protein (Bradley and Andrew

2001; Shim et al. 2001).Whereas most dorsal-open group mutants exhibit

striking defects only in dorsal closure, animals homozy-gous for recessive mutations in raw, puc, and rib die inembryogenesis with defects in dorsal closure and ventralcuticular patterning (Figure 3, B–D, F–H, and J–L).Each of the raw, puc, and rib gene products is requiredfor proper restriction of JNK signaling targets to LE cellsduring closure, as dpp and/or puc expression are shownin the current report and elsewhere to expand beyondthe LE in mutants (Blake et al. 1998; Byars et al. 1999;Bradley and Andrew 2001). Here we show that for allthree loci, the extent of ectopic signaling correlates withallele strength, and it is the null alleles that exhibit thebroadest signaling domains as measured by the expan-sion of dpp expression (Figure 3, A, E, and I). Whereasthe dpp expression phenotypes in puc and raw mutantsare comparable, there is considerably less expansion inthe rib mutant in comparison to either raw or puc mutantembryos (compare Figure 1B and Figure 3, A, E, and I;see also Bradley and Andrew 2001). The subset ofdorsal-open group genes sharing the raw palette of

phenotypes currently includes raw, puc, and rib, and isfrom hereon referred to as the raw group.

raw and puc function independently to antagonizeJNK/AP-1 activity: Although the raw-group genes sharean array of loss-of-function phenotypes, it remainedunclear whether these three genes (raw, puc, and rib) acttogether, in some combination, or independently torestrict JNK/AP-1 signaling to LE cells. The simplest andmost common signal attenuating regulatory mechanismin any biochemical network is negative feedback (forreview see Rao et al. 2002). Thus, we considered thepossibility that the raw and rib gene products functionwith Puckered in a single negative-feedback loop inac-tiving JNK via dephosphorylation and consequentlyinactivating AP-1 indirectly. As an alternative, we rea-soned that two or more AP-1 inactivation mechanismsmight function in parallel (and thereby combine) toeffect a robust signaling response (Kerszberg 2004).

We employed genetic interaction studies as an initialmeans to distinguish between linear and parallel rolesfor raw and puc in antagonizing the JNK/AP-1 pathway.Double-mutant animals harboring hypomorphic muta-tions can exhibit enhanced phenotypes when two genesfunction in either common or parallel pathways. Thus,our discovery that hypomorphic mutations in raw andpuc enhance one another’s cuticular phenotype wasnot unanticipated; both the dorsal closure and ventraldenticle defects are stronger in the raw2418; pucE69 doublemutants than in either mutant alone (Figure 4, A and B;see also Figure 1E). More notably, animals doubly mu-tant for null alleles will show evidence of enhancementonly when the two gene products function in parallelpathways. Here we show that raw1; pucH246 double mu-tants exhibit a very severe and fully penetrant cuticulardefect that is distinct from that of cuticles derived fromeither raw1 or pucH246 single-mutant embryos (Figure 4C;see also Figures 1A and 3F). These null-allele-interactiondata indicate that raw and puc function in independentand presumably parallel pathways to antagonize JNK/AP-1 signaling in Drosophila embryos.

Next, we turned our attention to the relationship be-tween raw and rib. In contrast to our discovery that nullalleles of raw and puc enhance one another’s phenotype,we found that null alleles of raw mask the weaker ribphenotype. The rawlex1 rib1 double-mutant dorsal-closuredefect is analogous to that of raw null homozygotes(Figure 4D; see also Figure 1A). Hence, while raw andpuc function independently to restrict JNK/AP-1 signal-ing in epidermal cells during closure, raw and rib (onthe basis of their epistatic relationship) likely functiontogether.

Raw could function to antagonize JNK/AP-1 at thelevel of transcription in the nucleus with Ribbon, or itcould antagonize AP-1 activity at the level of signaltransduction in the cytoplasm. To begin to distinguishbetween these possibilities, we examined the subcellu-lar localization of Raw. Two Raw fusion proteins, one

Figure 2.—Ventral denticle abnormalities in raw mutantembryos result from ectopic epidermal Dpp signaling. (A) Cu-ticles derived from raw1; UAS-brk/69BGAL4 embryos exhibitsignificant dorsal closure and head involution defects, compa-rable to those of the null homozygote shown in Figure 1A; theventral denticle belt phenotype, however, is largely rescued.(B and C) Magnification and trace drawing of ventral den-ticles in A. brk mRNA transcript expression in situ is shownin stage 13 whole-mount wild-type embryos (D) and raw1 mu-tant embryos (E). Whereas brk is expressed in lateral and ven-tral stripes in wild-type embryos, these expression domains aremissing in raw1 homozygotes.


harboring an N-terminal Myc tag, the other a C-terminalFLAG tag were expressed transiently in COS cells, andthe linked epitopes were detected using standardimmunological methods. In both cases, we found theRaw fusion protein to be cytoplasmic (Figure 5). Thetagged Raw isoforms are cytoplasmic in embryos as well;however, in contrast to untagged versions of Raw thatfunction in rescue assays in vivo (see Figure 6), thetagged isoforms do not (data not shown).

Together with the overlapping embryonic patterns ofraw and rib expression (Byars et al. 1999; Bradley andAndrew 2001; Shim et al. 2001), our genetic interactionstudies indicate that raw and rib function together indorsal closure as they do in several other developmentalcontexts (Blake et al. 1998, 1999). Nevertheless, the

Figure 3.—raw-group genes ex-hibit an array of shared loss-of-function phenotypes. (A, E, andI) dpp mRNA transcript expressionin whole mount embryos in situ;(B, F, and J) cuticular phenotypes;(C, G, and K) magnification of cu-ticular denticle phenotypes; and(D, H, and L) trace drawings ofdenticle magnifications. In wild-type embryos, dpp expression islimited to the single row of lead-ing-edge epidermal cells (A), andthe five ventral denticle rows areclearly evident (B–D). pucH246 ho-mozygotes exhibit expanded dppexpression in the lateral epidermis(E), a dorsal cuticular hole (F), andhypertrophied ventral denticlebelts that are similar to those inraw2418 hypomorphs (G and H).rib1 homozygotes exhibit expandeddpp expression as well as dorsal clo-sure( J)andventraldenticledefects(K and L).

Figure 4.—raw and puc function independently to antago-nize JNK/AP-1 signaling. (A) raw2418; pucE69 cuticles exhibitdorsal closure, head involution, and ventral denticle defectsthat define null alleles of raw (see also Figure 1A). (B) ThepucE69 cuticular phenotype is more subtle; it is characterizedby a dorsal pucker. (C) raw1; pucH246 cuticles exhibit a novelphenotype, which masks defects in dorsal closure diagnosticof cuticles derived from single-mutant null animals. (D)The rawlex1 rib1 cuticular phenotype is analogous to the rawlex1

(and raw1) null phenotype. Cuticle magnification is identicalin all panels.

Figure 5.—Raw is cytoplasmic. N-terminal Myc-tagged Raw(A–C) is cytoplasmic in COS-7 cells. (D–F) C-terminal FLAG-tagged Raw is cytoplasmic in COS-7 cells. (A) N-terminal Myc-tagged Raw was visualized using an anti-Myc antibody; (D)C-terminal FLAG-tagged Raw was visualized using an anti-FLAG.Nuclei were visualized by staining with DAPI (B and E). Mergedimages of fluorescent channels are shown (C and F).


gene products’ roles in mediating JNK/AP-1 antago-nism (at least in dorsal closure) appear to be separable,both in terms of their quantitatively distinct effects onsignaling (see Figure 3) and their qualitatively distinctsubcellular locales (see Figure 5). Also consistent withthis model of separable raw and rib gene functions isour inability to rescue the raw2418 and rib1 dorsal-closurephenotypes (n . 100 for each) with functional rib1

(Bradley and Andrew 2001) and raw1 (this study)transgenes, respectively.

Epidermal JNK/AP-1 regulation is partitioned spa-tially: Results from our genetic interaction studiespointed to a function for raw, independent of thePuckered MKP-mediated feedback loop that controlsBasket JNK activity cell autonomously. Thus, we consid-ered alternative roles for the raw gene product in mod-ulating JNK and/or AP-1 activity. As an example, Rawmight function upstream of the JNK cascade to regulatethe activity of an as yet undefined JNK activator (re-strictive signaling model); alternatively, Raw might func-tion in parallel to the JNK cascade to repress epidermalJNK and/or AP-1 activity (repression model) (Byars

et al. 1999).To distinguish between restrictive signaling and re-

pression models of raw function and to establish a reg-

ulatory link between raw and the genes encodingmembers of the JNK signaling cascade, we determinedtheir epistatic relationships. LE dpp expression is a veryclear indicator of this relationship since LE dpp tran-scription is missing in mutants for all JNK/AP-1 signal-ing components, but is greatly expanded in raw-groupmutants (Glise et al. 1995; Riesgo-Escovar et al. 1996;Glise and Noselli 1997; Su et al. 1998; Stronach andPerrimon 2002) (see also Figures 1 and 3). By examin-ing LE dpp expression in embryos doubly mutant formolecularly and genetically defined null alleles of rawand Jra (encoding the Jun component of AP-1), we haveshown that Jra is epistatic to raw and thus we identifiedan upstream role for raw as an antagonist of AP-1 activity(Byars et al. 1999).

We continued our analysis of raw’s epistatic relation-ships by examining raw’s relationship to basket (bsk),the gene encoding the upstream activating JNK ( Junkinase). For this analysis, we employed two geneticallydefined bsk null alleles (bsk1 and bsk2). The bsk1 and bsk2

lesions are distinct, affecting different conserved do-mains in the Basket JNK protein. bsk1 corresponds to amissense mutation (G225E) mapping between theBasket kinase subdomains IX and X. This region isrequired for JNK substrate recognition and is conservedamong members of the MAPK family to which theBasket JNK belongs. Molecularly, bsk1 is classified as a‘‘severe’’ allele; genetically, however, it is defined as null(Sluss et al. 1996), and it has been used previously inepistasis studies (Stronach and Perrimon 2002). Likebsk1, bsk2 corresponds to an EMS-induced mutation, inthis case targeting K316 and creating an in-frame stop.K316 (and its corresponding nonsense mutation) mapswithin the conserved kinase subdomain XI. Althoughnot demonstrated biochemically, the protein encodedby bsk2, like that encoded by bsk1, is predicted to beenzymatically inactive. Consistent with this predictionare the experimental observations that (1) in bsk2 em-bryos AP-1 activity is undetected in LE dpp expressionassays, and (2) the bsk2 embryonic lethal cuticular phe-notype is analogous to that of the genetically definedbsk1 and molecularly defined bskFLP17E null alleles (Riesgo-Escovar et al. 1996; Riesgo-Escovar and Hafen

1997a).Although both bsk1 and bsk2 mutant embryos lack all

dpp expression in LE epidermal cells, we found that dppis broadly expressed in the epidermis of both raw1 bsk1

and raw1 bsk2 double mutants (Figure 6, A and B, anddata not shown). Thus raw is epistatic to bsk, implyingthat raw functions either downstream of bsk or indepen-dently of bsk.

Together, our observations that (1) Jra is epistatic toraw and (2) raw is epistatic to bsk are consistent withmodels in which raw, bsk, and Jra function in a singlelinear pathway. When considered within this linearcontext, however, our epistasis results place raw squarelybetween bsk and Jra (Figure 6G), wedged within a well-

Figure 6.—raw-group genes share identical epistatic rela-tionships with genes encoding JNK-signaling molecules. dppmRNA transcript expression in whole-mount embryos in situ.(A) bsk1, (B) raw1 bsk1, (C) JraIA109; pucH246, (D) JraIA109; rib1, (E)bsk1; pucH246, and (F) bsk1; rib1. In A–F, the position of the lead-ing edge is marked by an arrow (Y). (G and H) Diagramsshowing alternative explanations for the epistatic relation-ships of JNK-signaling molecules to raw (and other raw-groupgenes) in leading-edge (LE) and epidermal (EP) cells moregenerally.


studied biochemical pathway that has, in all probability,been defined in its entirety. Indeed, Jun ( Jra) is a directtarget of Basket-mediated phosphorylation (for reviewsee Noselli and Agnes 1999). Genetically definedregulatory pathways are not, however, always linear;sometimes they are branched and/or contain multipleinputs. A more fitting interpretation of our results—thatJNK/AP-1 regulation is partitioned in a tissue-specificfashion during closure—is attained using a branchedpathway model instead of a linear model. More specif-ically, our data implicate Raw in epidermal repression ofAP-1 activity. While AP-1 activity in LE cells may bedependent upon Basket JNK activity, AP-1 activity inepidermal cells positioned more laterally is (1) Basketindependent and (2) repressed by Raw (Figure 6H).

With two additional genetically defined JNK/AP-1antagonists in hand, we next investigated whether theyexhibit similar or different epistatic relationships tocomponents of the JNK/AP-1 cascade. We examinedepidermal expression of dpp in Jra puc, Jra rib, bsk puc,and bsk rib double mutants (in all cases using geneticallyand/or molecularly defined nulls). We found thatwhereas dpp expression is absent in both Jra puc andJra rib double mutants (Figure 6, C and D), dpp ex-pression is expanded in both bsk puc and bsk rib doublemutants (Figure 6, E and F). Thus, the epistatic rela-tionships of rib and puc to the JNK/AP-1 cascade areidentical to that of raw and the JNK/AP-1 cascade. Withrespect to epidermal AP-1 activity, all three JNK/AP-1antagonists function upstream of Jra, but independentlyof the Basket JNK. Overall, these epistasis studies lead usto suggest that (1) the epidermis is broadly competentfor Jra/AP-1-mediated transcription, and (2) BasketJNK-mediated phosphorylation of epidermal Jra/AP-1appears to be dispensable for its activity as a transcrip-tion factor.

Raw functions broadly in the embryonic epidermisto establish an AP-1 silenced ground state: Our studiesto this point implicated a role for raw (and the raw-group genes more generally) in a previously uncharac-terized epidermal pathway(s) of AP-1 silencing. In adirect test of this model, we assessed the spatial require-ments for raw in dorsal closure. To this end, we es-tablished a raw transgenic rescue assay and employedtargeted expression of a UAS-raw1 transgene to deter-mine whether raw expression in any of the three tissuesparticipating in dorsal closure (amnioserosa, leadingedge epidermis or epidermis more generally) is suffi-cient to effect dorsal closure. Normally, raw is expressedin all three tissues (Byars et al. 1999).

We observed proper epithelial morphogenesis anddorsal closure in raw2418 homozygotes when we used the69B-GAL4 driver to express two independently derivedUAS-raw1 transgenes pan-epidermally. In contrast, wenever observed rescue of the raw2418 phenotypes whenwe used the Kr-GAL4 or LE-GAL4 drivers that targetUAS-raw1 gene expression to the amnioserosa or LE

epidermis, respectively (Figure 7A). These data indicatethat (1) expression of a raw1 transgene broadly in theepidermis is sufficient to rescue raw-dependent defectsin dorsal closure, and (2) expression of a raw1 trans-gene, when confined to either the amnioserosa or theLE epidermis, is not sufficient to rescue raw-dependentdefects in closure. These results, and those from ourepistasis studies, indicate that Raw mediates restrictionof AP-1 activity in a tissue-autonomous fashion, via theepidermis.

We next tested when expression of a heat-shock in-ducible raw1 transgene can rescue raw-dependent de-fects in dorsal closure. We found that when hs-raw1 (intwo independently derived lines) was induced beforethe start of dorsal closure at 4–8 hr AEL, rescue was veryefficient, with rates of 68 and 98%. In contrast, weobserved reduced rates of rescue (39 and 65%) when hs-raw1 transgenes were induced during dorsal closure,between 8 and 12 hr AEL (Figure 7B). These data in-dicate that Raw-mediated antagonism of AP-1 baselineactivity occurs relatively early in embryogenesis. More-over, these data are consistent with our hypothesis thatraw is required to establish an inactive epidermal AP-1ground state.

raw functions to antagonize AP-1 activity at multipledevelopmental time points: Having established that rawis an essential negative regulator of AP-1 activity inembryogenesis, we next tested whether raw is a dorsal

Figure 7.—Raw functions early in the epidermis to effect dor-sal closure. (A) Tissue-specific rescue in raw2418; p½UAS-raw1�/1transgenics. Rescue was observed only when raw expression wasdriven by the 69B-GAL4 epidermal driver. In this experimentalseries, we observed rescue of embryonic dorsal closure in 72%of raw2418 homozygotes. (B) Induction of either of two indepen-dently derived hs:raw1 transgenes in raw2418 homozygotes 4–8hr AEL rescues 68 and 98% of raw� homozygous embryos tothe first larval instar stage. Induction of the same transgenes8–12 hr AEL rescues 39 and 65% of raw� homozygotes. Inthe absence of heat shock, background rates of rescue rangefrom 9 to 25%.


closure-specific antagonist of AP-1 activity or whetherraw has a more general role throughout development inthe negative regulation of AP-1 prior to its activation byJNK-mediated signaling. Among the best characterizeddevelopmental and physiological requirements forJNK/AP-1 signaling in Drosophila are its roles in dorsalclosure (Glise et al. 1995; Riesgo-Escovar and Hafen

1997a,b), follicle cell morphogenesis (Dequier et al.2001; Dobens et al. 2001; Suzanne et al. 2001), andoxidative stress tolerance (Wang et al. 2003). Thus,there are three well-characterized JNK/AP-1-dependentprocesses in Drosophila, and we show here that eachone is affected in raw mutant animals.

During oogenesis, JNK/AP-1 signaling is required forfollicle cell (FC) morphogenesis. Reduced or elevatedlevels of JNK/AP-1 activity, achieved either by mutatingor overexpressing components of the JNK cascade insomatic FC clones result in a distinctive oocyte pheno-type. Signature defects in oocytes with improper levelsof JNK/AP-1 signaling include failures in nurse celldumping and dorsal appendage morphogenesis, in ad-dition to an overall reduction in oocyte length (Dobens

et al. 2001; Suzanne et al. 2001). raw might play a roleanalogous to puc in antagonizing JNK/AP-1 activityduring oogenesis as it is expressed in stretch cells (Fig-ure 8A), JNK/AP-1-active somatic FCs that overlie thenurse cells during oogenic stages 10B-11 and one ofthree migrating cell types in the Drosophila oocyte(Dequier et al. 2001).

For direct tests of raw function in oocytes, we used araw1 FRT40A chromosome in a FLP/FRT-mediatedclonal analysis (Xu and Rubin 1993). In this study ofraw1 mutant clones, we found that females with rawmutations in their germ line lay small, unfertilized eggs,many of which are distinguished by short dorsalappendages (Figure 8, B and C). We quantified the sizedefect and found that in contrast to eggs laid by wild-type females, 71% of which are at least 0.7 mm in length,only 22% of eggs laid by raw1 FRT40A females grow tothis same length. Indeed the majority (57%) is at least10% shorter (Figure 8E). We also tested whether rawoverexpression similarly leads to defects in oogenesis.Five hours after induction of a hs-raw1 transgene,females lay unfertilized eggs that are shorter than wild-type controls and exhibit dorsal appendage defects(Figure 8D). Quantization of length defects revealedthat 51% of eggs laid by females overexpressing rawexhibit at least a 10% reduction in length in comparisonto wild-type control animals (Figure 8F). Overall, theraw-associated loss- and gain-of-function defects mirrorthose observed in studies of puc, a phosphatase with amolecularly defined role in regulating JNK signaling.

Finally, we tested whether raw functions in oxidativestress tolerance. Animals with elevated levels of JNKsignaling and AP-1 activity, achieved either by over-expression of JNK-signaling components or by mutationof the puc-encoded MKP, exhibit an increased resistance

to oxidative stress as measured by life span (Wang et al.2003, 2005). We reasoned that if raw is a general an-tagonist of the JNK/AP-1 signaling pathway, thenelevated levels of AP-1 activity in raw mutants will affectoxidative stress tolerance and aging in a manneranalogous to that of puc. To test this hypothesis, weexposed raw and puc adult heterozygotes to oxidativestress in the form of paraquat treatment and measuredlife span. In comparisons of mortality rates in 1/1 ho-mozygotes, puc/1, and raw/1 heterozygotes, we foundthat as reported previously puc heterozygotes exhibitdecreased mortality levels. Notably, we found that areduction in raw also leads to a mortality decrease (Fig-ure 8G), consistent with raw functioning as an antago-nist of JNK signaling during the adult response tooxidative stress. To control for effects of genetic back-ground, we employed two independently derived nullalleles of raw. As was the case for puc, the raw heterozy-gote effect of life span demonstrates a dosage-sensitiverequirement for raw in oxidative stress tolerance. Takentogether, results from this and the oogenesis study pointto a role for raw as a negative regulator of AP-1 activitythroughout the Drosophila life cycle.

DISCUSSION

Our initial molecular and genetic studies of thedorsal-open mutant raw revealed it to encode a widelyexpressed and novel gene product, required for therestriction of JNK/AP-1 activity to LE epidermal cells(Byars et al. 1999). The Raw protein sequence yieldedno insights into its mechanism of function as the Rawsequence harbors none of the canonical motifs that areassociated with nuclear localization, phosphorylation,membrane insertion, or protein secretion. Mechanisticstudies of a novel protein can be challenging, but herewe report our use of a variety of genetic strategies toprobe Raw function and test models of AP-1 silencing.In particular we (1) assessed the epistatic relationship ofraw to genes encoding well-characterized JNK-signalingcomponents, (2) identified genes, which we have des-ignated the raw group, that share an array of loss-of-function phenotypes, (3) determined the interactionphenotypes among the raw-group loci, and (4) gener-ated raw transgenics, which we utilized to probe sites ofRaw function. Our analyses reveal that raw belongs to asmall set of dorsal-open group genes that encode JNK/AP-1 pathway antagonists. Our characterization of raw,and the raw group more generally, has led to a newappreciation of wide-ranging competence for AP-1activity in early Drosophila embryos. As signal activationis critical for proper development, so also is its silencing.

Regulating AP-1 baselines in Drosophila: In the cur-rent study, we show that although raw functions up-stream of Jra as an AP-1 antagonist, its action isindependent of the bsk-encoded kinase that is required


to activate AP-1 activity in LE cells during closure. Inaddition, we demonstrate that raw is required broadly inthe epidermis to effect normal dorsal closure. Overall,our studies expose the importance of epidermal AP-1silencing during embryogenesis and lead us to extend

Figure 9.—Modeling Raw function. (A) The canonical JNKcascade is active in LE cells and presumably controlled by thenegative-feedback regulator Puckered. In the adjacent epider-mal (EP) cells, Basket-independent AP-1 activity is silenced byRaw. (B) Baseline/basal levels of AP-1 activity are kept belowbiologically relevant levels by the actions of Raw, Ribbon,and Puckered. Activation of the JNK-signaling cascade inthe leading-edge cells supersedes raw-group gene product-mediated AP-1 silencing. Lastly, biologically appropriate sig-naling levels are thought to be maintained via the activityof the JNK negative-feedback regulator Puckered.

Figure 8.—raw antagonizes JNK/AP-1 signaling at multipleDrosophila life stages. (A) raw mRNA transcripts are ex-pressed in stretch cells; for this analysis oocytes were fixedand hybridized with an antisense RNA probe directed againstraw. (B–D) raw-dependent gain- and loss-of-function egg phe-notypes include size and dorsal appendage defects (indicatedby arrows,Y). (B) 1/1, (C) hs-raw1, and (D) raw1 follicle cellclones. (E and F) Quantization of gain- and loss-of-function-size phenotypes. (E) Eggs derived from wild-type females(dotted lines), exposed to either heat shock (¤) or not()), are rarely smaller than 0.63 mm. In similar fashion, eggsderived from females harboring a raw1 transgene that has notbeen induced by heat shock (solid lines) are also only infre-quently smaller than 0.63 mm (n). In contrast, a significant

fraction of eggs (34.8%) derived from transgenic females af-ter raw1 induction by heat shock (:). (F) Eggs derived fromfemales harboring raw1 follicle cell clones exhibit a widerrange of sizes than do controls. Twenty-one percent of eggsproduced by FLP/FRT raw1 females (solid line) are smallerthan 0.63 mm, whereas only 1.1% of eggshells produced bywild-type control females (dotted line) are smaller than0.63 mm. (G) Reducing JNK/AP-1 signaling extends life spanafter exposure to oxidative stress. Wild-type adult femaleshave an average lethality of 22.5% 24 hr after exposure toparaquat. Removal of one copy of puc reduces lethality to2.0%. Removal of one copy of raw, using null (raw1 and rawlex1)or hypomorphic (raw2418) alleles reduces the lethality to 4.5and 5.0%, respectively.


existing models for dorsal closure, which have largelyconfined their focus to mechanisms of JNK/AP-1activation in LE cells (Young et al. 1993; Martin-Blanco et al. 1998; Noselli 1998). In particular, ourdata indicate that Raw and the other raw-group geneproducts (Puckered and Ribbon) function to silenceBasket JNK-independent AP-1 activity in the embryonicdorsolateral epidermis (Figure 9A). AP-1 silencing, viathe combined actions of the raw-group gene products,essentially wipes the epidermal slate clean and primesthe system for activation via a still unidentified de-terministic signal that acts only in LE cells (Figure 9B).

We have yet to molecularly define the AP-1 abnor-mality in raw-group mutant embryos. Previous studiesprovide compelling evidence that AP-1 overexpressionin Drosophila embryos is not sufficient to disrupt eitherdorsal closure or development more generally (Kockel

et al. 1997; Zeitlinger et al. 1997). It seems unlikely,therefore, that elevated levels of the AP-1 transcriptionfactor in raw-group mutants simply override a require-ment for kinase activation in initiating an AP-1-dependent program of gene expression. Instead, wespeculate that AP-1 is aberrantly modified in raw-groupmutant embryos. It might be that AP-1 escapes inac-tivation in mutants; either alternatively or additionally,AP-1 in mutants may be inappropriately activated viaphosphorylation. In addition to Basket JNK, there arefour other Drosophila MAP Kinases (p38a, p38b, Mpk2,and Rolled) that might provide dysregulated kinaseactivity in mutants. Consistent with this idea is our ob-servation that the oogenesis phenotypes associated withraw (and puc) ectopic expression and mutation haveconsiderable similarity with gain- and loss-of-functionphenotypes associated with mutations in the p38 path-way that is required in the germ line for properoogenesis (Suzanne et al. 1999). Finally, a kinase-dependent activation model for epidermal Jun providesthe most parsimonious explanation for ectopic epider-mal signaling observed in puc MPK-deficient embryos.From the perspective of regulated signaling moregenerally, however, lowering an AP-1 activity baselinein wild-type embryos will (1) provide a means for theclean on/off regulation of JNK/AP-1 that has been pre-dicted in computer simulations and (2) make a lessstrenuous demand on the input activating signal (Sauro

2004) (Figure 9B).The different contributions of Raw, Ribbon, and

Puckered to JNK/AP-1 regulation: Our discovery thatnull alleles of raw and puc interact, with double mutantsexhibiting an embryonic lethal phenotype distinct fromtheir shared loss-of-function null phenotypes, revealedthe independent contributions of raw and puc to em-bryogenesis, presumably through their effects on AP-1antagonism. Drosophila overexpression studies havepreviously implicated several pathways in the parallelcontrol of AP-1 activity (Gritzan et al. 2002; Kozlova

and Thummel 2003), but our analysis represents the

first direct demonstration of physiologically relevant,parallel regulatory pathways.

The genetic interaction that we documented betweennull alleles of raw and puc contrasts with the lack of adetectable interaction between null alleles of raw andrib. Moreover, our observation that raw and rib hypo-morphs interact genetically during dorsal closure (datanot shown) is consistent with previously published data,as well as with findings documenting (1) raw/rib inter-actions in several other epithelial tissues, including thenervous system, salivary gland, trachea, and gut (Blake

et al. 1998, 1999) and (2) overlapping raw and rib ex-pression patterns in Drosophila embryos (Byars et al.1999; Bradley and Andrew 2001; Shim et al. 2001).Together, results from these genetic and molecular stud-ies point to roles for raw and rib in a single, previouslyunrecognized puc-independent AP-1 inactivation system.

Activating JNK signaling in the LE: In addition toproviding evidence for raw-mediated global silencing ofAP-1, our study underscores a simultaneous require-ment for a biologically appropriate activator of JNK/AP-1 signaling. In this regard, expression of raw in LE cellsfailed to rescue raw-dependent defects in dorsal closure.Even more notable, however, was our observation thatoverexpression of raw1 in wild-type embryos, and inwild-type LE cells in particular, had no detrimentaleffects on embryonic development and dorsal closure.From a signaling perspective this result indicates thatJNK-dependent AP-1 can be activated despite expres-sion of the wild-type raw gene product, and thus Rawdoes not function as a binary switch for signaling.Although it is formally possible that LE expression ofraw was initiated too late to disrupt JNK/AP-1 signalingand dorsal closure in our LE-gal4/UAS-raw1 transgenics,we do not favor this interpretation as the LE-GAL4driver used here has been shown previously to (1) be aneffective driver of at least one gene that is required in LEepidermal cells for closure (Lu and Settleman 1999)and (2) drive expression of a lacZ reporter in LE cellsduring dorsal closure (Scuderi and Letsou 2005).

Our finding that raw expression in LE cells is notsufficient to inactivate AP-1 activity in a cell-autonomousfashion is consistent with models for independent,developmentally regulated triggers of JNK signaling.Indeed, there is abundant experimental support fordevelopmentally regulated activation of JNK signalingin LE cells. JNK/AP-1 activation likely follows an amal-gamation of signals, both from the amnioserosa and theepidermis, both in the form of cytoskeletal componentsand signaling molecules (Harden 2002; Scuderi andLetsou 2005). Among the best candidates with postu-lated roles in JNK/AP-1 activation are small GTPases,nonreceptor tyrosine kinases, and integrins (Glise et al.1995; Stark et al. 1997; Kiyokawa et al. 1998; Bishop

and Hall 2000; Brown et al. 2000; Fernandez et al.2000). Thus, despite the broad epidermal competencefor AP-1 signaling that we have shown in this work, the


activation signal is itself limited to only LE cells andfunctions via an unknown mechanism. Importantly,AP-1 antagonism by raw cannot override its signal-dependent activation in the LE.

Activating brinker in the ventral epidermis: dpp, whenexpressed pan-epidermally, leads to a raw-like pheno-type: embryonic lethality associated with ventral cuticu-lar defects (Riesgo-Escovar and Hafen 1997a; Byars

et al. 1999). In a direct assessment of equivalence of rawloss-of-function and dpp gain-of-function ventral cutic-ular phenotypes, we tested whether pan-epidermalexpression of brinker (brk) can rescue raw-dependentdefects in the ventral cuticle. The Dpp signaling mod-ifier Brinker functions by negatively regulating dpptarget genes (Bray 1999; Campbell and Tomlinson

1999; Jazwinska et al. 1999a; Minami et al. 1999).In our analysis, we found that although brk is normally

expressed in nonoverlapping lateral and ventral do-mains of the embryonic epidermis, it is undetectable inthe epidermis of embryos homozygous for a null alleleof raw. We also found that although brk1 fails to rescueraw-dependent defects in dorsal closure, it does rescueraw-dependent defects in the ventral cuticle. Together,these data point to an important role for dpp, brk, and/or their target genes in development of the ventralepidermis.

What we cannot discern from our studies is (1) howthe nonoverlapping epidermal domains of dpp and brkare established and maintained and (2) if and howepidermal dpp and brk interact during normal embry-onic development. In this regard, our previous findingthat LE dpp is not autoregulatory makes it unlikely thatbrk functions in direct fashion to set the LE dppexpression boundary (Arora et al. 1995). Even moresignificant is our finding that cuticles derived from Jraraw double mutants exhibit defects in dorsal closure,but not ventral cuticular patterning (Byars et al. 1999).Indeed, these data highlight the requirement forfunctional Jun in generating ventral cuticular defectsin raw mutant embryos. Taken together then, our datasuggest that the effects of JNK/AP-1-activated dpp in thedorsal epidermis of raw mutant embryos are far reach-ing, extending even to the most ventral regions of theembryo.

Having established a dependence upon Jun for raw-dependent ventral cuticular defects, we postulate thatthe absence of brk in raw mutant embryos is a directconsequence of ectopic JNK/AP-1 activity in the dorsalepidermis of these mutants. We suspect that ectopicJNK/AP-1 activity leads secondarily to ectopic dpp ac-tivity, and that in its turn ectopic dpp activity leads finallyto brk repression. An alternative view, that raw mighthave dual regulatory roles in the epidermis, seems lesslikely although it is not absolutely excluded by ourstrictly genetic analysis. In this regard, in addition to itsfunction as a JNK/AP-1 antagonist in the embryonicdorsal epidermis, raw might function independently as

a trigger of brk expression in the ventral epidermis.Clearly, the mechanism of raw function and the re-lationship of dpp to brk in eliciting properly formedventral cuticle warrant further investigation.

Conclusions and implications for the study of JNK/AP-1 antagonists: In Drosophila, as in all animals, sig-naling pathways are finely regulated at several levels.Although there are multiple tiers of regulation operat-ing on the JNK/AP-1 signaling cascade, surprisinglylittle of the regulation of this pathway is known. Ourstudy of the functions and interactions of a subset ofdorsal-open group genes (raw, rib, and puc) has shedsome additional light on both old (puc-mediated) andnew (raw/rib-mediated) mechanisms of JNK/AP-1 an-tagonism. Our data indicate that Raw functions tosilence Basket JNK-independent AP-1-mediated tran-scription and to set the stage for JNK-dependent reg-ulation of transcription. Our suggestion that spatialrestriction of the JNK/AP-1 signal requires antagonists,as well as activators, is not without precedent in othersignaling systems. Many signaling pathways have alreadybeen shown to be multilayered and to depend heavily onnegative regulation to terminate developmental events,and/or control both the distance and speed that a signalcan move (e.g., Nodal; Shen 2007). In addition, and aswe suggest is the case for the Drosophila JNK/AP-1pathway, reducing basal levels of a signaling pathway canaugment the effects of its signaling responses (e.g.,Hedgehog and Lef1; Li et al. 2006; Flynt et al. 2007).

Finally, given the numerous associations of improperJNK/AP-1 activity with human disease, it seems appar-ent that many cell types have the capacity to signal viathe JNK/AP-1 pathway. Presumably, this capacity isdiminished (and then tightly regulated) during normalvertebrate development and aging. Viewed from thisperspective, our characterization of Raw as an essentialAP-1 antagonist establishes a clear basis for futurestudies of AP-1 regulation.

We thank Cherie Byars and other members of our laboratory forvaluable discussions. We also thank Susan Mango and Steve Wassermanfor comments on the manuscript, Diana Lim for figure preparation, andDeborah Andrew, Kendal Broadie, Michelle Lamka, Stephan Noselli,and Beth Stronach for fly lines. This work was supported by NationalInstitutes of Health grant R01GM-68083 to A.L.

LITERATURE CITED

Arora, K., H. Dai, S. Kazuko, J. Jamal, M. O’Connor et al.,1995 The Drosophila schnurri gene acts in the Dpp/TGF betasignaling pathway and encodes a transcription factor homolo-gous to the human MBP family. Cell 81: 781–790.

Bishop, A. L., and A. Hall, 2000 Rho GTPases and their effectorproteins. Biochem. J. 348(Pt 2): 241–255.

Blake, K. J., G. Myette and J. Jack, 1998 The products of ribbonand raw are necessary for proper cell shape and cellular localiza-tion of nonmuscle myosin in Drosophila. Dev. Biol. 203: 177–188.

Blake, K. J., G. Myette and J. Jack, 1999 ribbon, raw, and zipperhave distinct functions in reshaping the Drosophila cytoskeleton.Dev. Genes Evol. 209: 555–559.


Bradley, P. L., and D. J. Andrew, 2001 ribbon encodes a novelBTB/POZ protein required for directed cell migration in Dro-sophila melanogaster. Development 128: 3001–3015.

Bray, S., 1999 DPP on the brinker. Trends Genet. 15: 140.Brown, N. H., S. L. Gregory and M. D. Martin-Bermudo,

2000 Integrins as mediators of morphogenesis in Drosophila.Dev. Biol. 223: 1–16.

Byars, C. L., K. L. Bates and A. Letsou, 1999 The dorsal-opengroup gene raw is required for restricted DJNK signaling duringclosure. Development 126: 4913–4923.

Campbell, G., and A. Tomlinson, 1999 Transducing the Dpp mor-phogen gradient in the wing of Drosophila: regulation of Dpptargets by brinker. Cell 96: 553–562.

Debry, P., E. A. Nash, D. W. Neklason and J. E. Metherall,1997 Role of multidrug resistance P-glycoproteins in choles-terol esterification. J. Biol. Chem. 272: 1026–1031.

Dequier, E., S. Souid, M. Pal, P. Maroy, J. A. Lepesant et al.,2001 Top-DER- and Dpp-dependent requirements for the Dro-sophila fos/kayak gene in follicular epithelium morphogenesis.Mech. Dev. 106: 47–60.

Dobens, L. L., E. Martin-Blanco, A. Martinez-Arias, F. C. Kafatos

and L. A. Raftery, 2001 Drosophila puckered regulates Fos/Jun levels during follicle cell morphogenesis. Development128: 1845–1856.

Fernandez, R., F. Takahashi, Z. Liu, R. Steward, D. Stein et al.,2000 The Drosophila shark tyrosine kinase is required for em-bryonic dorsal closure. Genes Dev. 14: 604–614.

FlyBase Consortium, 2003 The FlyBase database of the Drosoph-ila genome projects and community literature. Nucleic Acids Res.31: 172–175.

Flynt, A. S., N. Li, E. J. Thatcher, L. Solnica-Krezel and J. G.Patton, 2007 Zebrafish miR-214 modulates Hedgehog signal-ing to specify muscle cell fate. Nat. Genet. 39: 259–263.

Glise, B., H. Bourbon and S. Noselli, 1995 hemipterous encodesa novel Drosophila MAP kinase kinase, required for epithelialcell sheet movement. Cell 83: 451–461.

Glise, B., and S. Noselli, 1997 Coupling of Jun amino-terminal ki-nase and Decapentaplegic signaling pathways in Drosophila mor-phogenesis. Genes Dev 11: 1738–1747.

Gritzan, U., C. Weiss, J. Brennecke and D. Bohmann, 2002 Trans-repression of AP-1 by nuclear receptors in Drosophila. Mech.Dev. 115: 91–100.

Harden, N., 2002 Signaling pathways directing the movement andfusion of epithelial sheets: lessons from dorsal closure in Dro-sophila. Differentiation 70: 181–203.

Hasty, J., J. Pradines, M. Dolnik and J. J. Collins, 2000 Noise-based switches and amplifiers for gene expression. Proc. Natl.Acad. Sci. USA 97: 2075–2080.

Huang, L. S., and P. W. Sternberg, 1995 Genetic dissection of de-velopmental pathways. Methods Cell Biol. 48: 97–122.

Jazwinska, A., N. Kirov, E. Wieschaus, S. Roth and C. Rushlow,1999a The Drosophila gene brinker reveals a novel mechanismof Dpp target gene regulation. Cell 96: 563–573.

Jazwinska, A., C. Rushlow and S. Roth, 1999b The role of brinkerin mediating the graded response to Dpp in early Drosophila em-bryos. Development 126: 3323–3334.

Kerszberg, M., 2004 Noise, delays, robustness, canalization and allthat. Curr. Opin. Genet. Dev. 14: 440–445.

Keyse, S. M., and M. Ginsburg, 1993 Amino acid sequence similar-ity between CL100, a dual-specificity MAP kinase phosphataseand cdc25. Trends Biochem. Sci. 18: 377–378.

Kiyokawa, E., Y. Hashimoto, T. Kurata, H. Sugimura and M.Matsuda, 1998 Evidence that DOCK180 up-regulates signalsfrom the CrkII-p130(Cas) complex. J. Biol. Chem. 273: 24479–24484.

Knust, E., 1997 Drosophila morphogenesis: movements behind theedge. Curr. Biol. 7: R558–R561.

Kockel, L., J. Zeitlinger, L. Staszewski, M. Mlodzik and D.Bohmann, 1997 Jun in Drosophila development: redundantand nonredundant functions and regulation by two MAPK signaltransduction pathways. Genes Dev. 11: 1748–1758.

Kozlova, T., and C. S. Thummel, 2003 Essential roles for ecdysonesignaling during Drosophila mid-embryonic development. Sci-ence 301: 1911–1914.

Lemischka, I. R., 2005 Stem cell biology: a view toward the future.Ann. NY Acad. Sci. 1044: 132–138.

Li, T. W., J. H. Ting, N. N. Yokoyama, A. Bernstein, M. van de

Wetering et al., 2006 Wnt activation and alternative promoterrepression of LEF1 in colon cancer. Mol. Cell. Biol. 26: 5284–5299.

Lu, Y., and J. Settleman, 1999 The Drosophila Pkn protein kinaseis a Rho/Rac effector target required for dorsal closure duringembryogenesis. Genes Dev. 13: 1168–1180.

Martin-Blanco, E., A. Gampel, J. Ring, K. Virdee, N. Kirov et al.,1998 puckered encodes a phosphatase that mediates a feed-back loop regulating JNK activity during dorsal closure in Dro-sophila. Genes Dev. 12: 557–570.

Minami, M., N. Kinoshita, Y. Kamoshida, H. Tanimoto and T.Tabata, 1999 brinker is a target of Dpp in Drosophila thatnegatively regulates Dpp-dependent genes. Nature 398: 242–246.

Muda, M., A. Theodosiou, N. Rodrigues, U. Boschert, M. Camps

et al., 1996 The dual specificity phosphatases M3/6 and MKP-3are highly selective for inactivation of distinct mitogen-activatedprotein kinases. J. Biol. Chem. 271: 27205–27208.

Noselli, S., 1998 JNK signaling and morphogenesis in Drosophila.Trends Genet. 14: 33–38.

Noselli, S., and F. Agnes, 1999 Roles of the JNK signaling pathwayin Drosophila morphogenesis. Curr. Opin. Genet. Dev. 9: 466–472.

Qi, M., and E. A. Elion, 2005 MAP kinase pathways. J. Cell Sci. 118:3569–3572.

Rao, C. V., D. M. Wolf and A. P. Arkin, 2002 Control, exploitationand tolerance of intracellular noise. Nature 420: 231–237.

Riesgo-Escovar, J., and E. Hafen, 1997a Drosophila Jun kinaseregulates expression of decapentaplegic via the ETS-domain pro-tein Aop and the AP-1 transcription factor DJun during dorsalclosure. Genes Dev. 11: 1717–1727.

Riesgo-Escovar, J. R., and E. Hafen, 1997b Common and distinctroles of DFos and DJun during Drosophila development. Science278: 669–672.

Riesgo-Escovar, J., M. Jenni, A. Fritz and E. Hafen, 1996 TheDrosophila Jun-N-terminal kinase is required for cell morpho-genesis but not for DJun-dependent cell fate specification inthe eye. Genes Dev. 10: 2759–2768.

Sauro, H., 2004 The computational versatility of proteomic signal-ing networks. Curr. Proteomics 1: 67–81.

Scuderi, A., and A. Letsou, 2005 Amnioserosa is required for dor-sal closure in Drosophila. Dev. Dyn. 232: 791–800.

Shen, M. M., 2007 Nodal signaling: developmental roles and regu-lation. Development 134: 1023–1034.

Shim, K., K. J. Blake, J. Jack and M. A. Krasnow, 2001 The Dro-sophila ribbon gene encodes a nuclear BTB domain protein thatpromotes epithelial migration and morphogenesis. Develop-ment 128: 4923–4933.

Sluss, H., Z. Han, T. Barrett, R. Davis and Y. Ip, 1996 A JNK signaltransduction pathway that mediates morphogenesis and an im-mune response in Drosophila. Genes Dev. 10: 2745–2758.

Staehling-Hampton, K., and F. Hoffmann, 1994 Ectopic decapen-taplegic in the Drosophila midgut alters the expression of fivehomeotic genes, dpp, and wingless, causing specific morpholog-ical defects. Dev. Biol. 164: 502–512.

Stark, K. A., G. H. Yee, C. E. Roote, E. L. Williams, S. Zusman et al.,1997 A novel alpha integrin subunit associates with betaPS andfunctions in tissue morphogenesis and movement during Dro-sophila development. Development 124: 4583–4594.

Stronach, B., 2005 Dissecting JNK signaling, one KKKinase at atime. Dev. Dyn. 232: 575–584.

Stronach, B., and N. Perrimon, 2002 Activation of the JNK path-way during dorsal closure in Drosophila requires the mixed lin-eage kinase, slipper. Genes Dev. 16: 377–387.

Su, Y., J. Treisman and E. Skolnik, 1998 The Drosophila Ste20-related kinase misshapen is required for embryonic dorsal clo-sure and acts through a JNK MAPK module on an evolutionarilyconserved signaling pathway. Genes Dev. 12: 2371–2380.

Suzanne, M., K. Irie, B. Glise, F. Agnes, E. Mori et al., 1999 TheDrosophila p38 MAPK pathway is required during oogenesisfor egg asymmetric development. Genes Dev. 13: 1464–1474.


Suzanne, M., N. Perrimon and S. Noselli, 2001 The DrosophilaJNK pathway controls the morphogenesis of the egg dorsal ap-pendages and micropyle. Dev. Biol. 237: 282–294.

Tanoue, T., M. Adachi, T. Moriguchi and E. Nishida, 2000 A con-served docking motif in MAP kinases common to substrates, ac-tivators and regulators. Nat. Cell Biol. 2: 110–116.

Wang, M. C., D. Bohmann and H. Jasper, 2003 JNK signaling con-fers tolerance to oxidative stress and extends lifespan in Dro-sophila. Dev. Cell 5: 811–816.

Wang, M. C., D. Bohmann and H. Jasper, 2005 JNK extends lifespan and limits growth by antagonizing cellular and organism-wide responses to insulin signaling. Cell 121: 115–125.

Xu, T., and G. M. Rubin, 1993 Analysis of genetic mosaics in devel-oping and adult Drosophila tissues. Development 117: 1223–1237.

Young, P., A. Richman, A. Ketchum and D. Kiehart, 1993 Mor-phogenesis in Drosophila requires nonmuscle myosin heavychain function. Genes Dev. 7: 29–41.

Zeitlinger, J., L. Kockel, F. A. Peverali, D. B. Jackson and D.Bohmann, 1997 Defective dorsal closure and loss of epidermaldecapentaplegic expression in Drosophila fos mutants. EMBO J.16: 7393–7401.

Communicating editor: T. Schupbach


raw mediates antagonism of ap-1 activity in drosophila mediates antagonism of ap-1 activity in...

Documents