Copyright � 2007 by the Genetics Society of AmericaDOI: 10.1534/genetics.106.067488

The Nonmuscle Myosin Phosphatase PP1b (flapwing) Negatively RegulatesJun N-Terminal Kinase in Wing Imaginal Discs of Drosophila

Jasmin Kirchner,* Sascha Gross,*,1 Daimark Bennett* and Luke Alphey*,†,2

*Department of Zoology, University of Oxford, Oxford OX1 3PS, United Kingdom and †Oxitec, Oxford OX14 4RX, United Kingdom

Manuscript received October 30, 2006Accepted for publication January 17, 2007

ABSTRACT

Drosophila flapwing (flw) codes for serine/threonine protein phosphatase type 1b (PP1b). Regulation ofnonmuscle myosin activity is the single essential flw function that is nonredundant with the three closelyrelated PP1a genes. Flw is thought to dephosphorylate the nonmuscle myosin regulatory light chain,Spaghetti Squash (Sqh); this inactivates the nonmuscle myosin heavy chain, Zipper (Zip). Thus, strong flwmutants lead to hyperphosphorylation of Sqh and hyperactivation of nonmuscle myosin activity. Here, weshow genetically that a Jun N-terminal kinase (JNK) mutant suppresses the semilethality of a strong flw allele.Alleles of the JNK phosphatase puckered (puc) genetically enhance the weak allele flw 1, leading to severe wingdefects. Introducing a mutant of the nonmuscle myosin-binding subunit (Mbs) further enhances thisgenetic interaction to lethality. We show that puc expression is upregulated in wing imaginal discs mutant forflw 1 and pucA251 and that this upregulation is modified by JNK and Zip. The level of phosphorylated (active)JNK is elevated in flw 1 enhanced by puc. Together, we show that disruption of nonmuscle myosin activatesJNK and puc expression in wing imaginal discs.

PROTEIN phosphatase type 1 (PP1), a major class ofserine/threonine protein phosphatases, is one ofthe most conserved proteins in the animal kingdomand has been found in all eukaryotes examined to date(Lin 1999). PP1 is involved in the regulation of manycellular functions, including glycogen mechanism, mus-cle contraction, and mitosis (reviewed in Bollen 2001;Cohen 2002). In vitro, the catalytic subunit of PP1 (PP1c)dephosphorylates a wide variety of substrates, while in vivoPP1c is bound to a number of different regulators thatmodify its substrate activity and specificity and target it tospecific locations.

Drosophila melanogaster has four genes encoding PP1c.Three code for the PP1a type (Pp1-13C, Pp1-87B, andPP1a-96A), while one gene, flapwing ( flw, PP1b9C ),codes for the PP1b type. Of these genes, Pp1-87B is themajor PP1 form and contributes �80% of the total PP1activity in third instar larvae (Dombrádi et al. 1990).Although the amino acid sequences of PP1a and PP1bare extremely similar, they show structural differencesthat are conserved in vertebrates; i.e., Drosophila PP1a ishomologous to mammalian PP1a and PP1g, and PP1b ishomologous to mammalian PP1b/d. This suggests a con-served functional difference between the two subtypes.

We previously showed that flw has a single essentialfunction as a nonmuscle myosin phosphatase that isnonredundant with PP1a. Flw, but not PP1a, bindsspecifically to the myosin phosphatase-targeting sub-unit MYPT-75D, the homologue of mammalian MYPT3(Vereshchagina 2004). Recently, phosphorylation ofhuman MYPT3 was shown to activate MYPT3-associatedPP1d toward cytoplasmic myosin regulatory light chain(Yong 2006). In Drosophila, MYPT-75D targets Flw tononmuscle myosin, where it presumably dephosphory-lates the nonmuscle myosin regulatory light chain,Spaghetti squash (Sqh), thereby deactivating nonmusclemyosin. In clones mutant for the strong allele flw6, thelevel of phosphorylated Sqh (P-Sqh) is elevated and thenonmuscle myosin heavy chain Zipper (Zip) is hyper-activated. Reducing the amount of active Zip suppressesthe semilethality of flw6, indicating that regulation ofnonmuscle myosin activity is the only essential functionof flw. Conversely, expressing a constitutively active formof Sqh enhances the weak viable allele flw 1 towardcomplete lethality (Vereshchagina 2004).

Another myosin phosphatase-targeting subunit inDrosophila is the myosin-binding subunit (Mbs), whichis the homologue of mammalian MYPT1/2; it bindsboth Flw and Pp1-87B in vitro (Vereshchagina 2004).The level of P-Sqh is elevated in Mbs mutant clones (Leeand Treisman 2004). This suggests that Mbs targetsPP1c to its substrate Sqh, and therefore flw might not bethe only Sqh phosphatase. However, it is likely that thetwo presumptive nonmuscle myosin phosphatases Mbs/PP1c and MYPT-75D/Flw are localized differentially in

1Present address: Abbott Laboratories Global Pharmaceutical RegulatoryAffairs, Abbott Park, IL 60064-6157.

2Corresponding author: Department of Zoology, University of Oxford,South Parks Rd., Oxford OX1 3PS, United Kingdom.E-mail: luke.alphey@zoo.ox.ac.uk

Genetics 175: 1741–1749 (April 2007)

cells because MYPT-75D has a prenylation motif whileMbs does not (Vereshchagina 2004). This, togetherwith the specificity of MYPT-75D for Flw, may accountfor the evident lack of redundancy between Flw andother potential Sqh phosphatases.

Drosophila Jun N-terminal kinase (DJNK, basket, bsk) isa mitogen-activated protein kinase (MAPK) that regu-lates several essential developmental processes in Droso-phila (dorsal closure, thorax closure, dorsal appendageformation) as well as wound healing (Martin-Blancoet al. 1998; Zeitlinger 1999; Suzanne 2001; Ramet et al.2002). JNK is phosphorylated and activated by the MAPKkinase hemipterous (Hep) and phosphorylates Jun andFos, which together form the active transcription factorAP-1. AP-1 induces the transcription of a number ofgenes, among them puckered (puc), which encodes a JNKphosphatase and thus provides a negative feedback loop(reviewed in Kockel 2001).

Some target genes of JNK-activated AP-1 in dorsalclosure are components of the actomyosin network, forexample, Drosophila profilin (chickadee), Tropomyosin1(Tm1), Tropomyosin2 (Tm2), and myosin light chain 2(mlc-2) ( Jasper 2001). Nonmuscle myosin also has anessential role in dorsal closure (shown for zip, sqh, andMbs) (Young 1993; Jordan 1995; Mizuno 2002).

Even though regulation of nonmuscle myosin andJNK is crucial for such processes as dorsal closure orwound healing, no genetic or molecular interactionbetween the two has been found so far. In this study, weshow that the nonmuscle myosin phosphatase flwgenetically interacts with members of the JNK pathway.Mutants in JNK (bsk) suppress the strong semilethalallele flw6, while expression of hepCA, a hyperactive formof the JNK kinase, enhances the weak viable allele flw 1.Hypomorphic mutants in the JNK phosphatase pucalso enhance flw 1, leading to severe wing defects. Thisgenetic interaction is further enhanced by a mutant inthe myosin-binding subunit Mbs. We show that flw neg-atively regulates puc expression in wing imaginal discsand that this regulatory action is modulated by JNKand the nonmuscle myosin heavy chain Zip. Finally, thelevel of diphosphorylated (active) JNK is elevated inprotein extracts from flw 1/ Y ; puc A251/1 wing imaginaldiscs. Ectopic activation of JNK in wing imaginal discs isknown to induce apoptosis (Adachi-Yamada et al.1999a), suggesting that the wing phenotype exhibitedby flw 1/Y ; puc A251/1 flies is due to increased cell death.Together, we show that flw can activate JNK through afunction in the regulation of nonmuscle myosin.

MATERIALS AND METHODS

Fly genetics and wing preparations: D. melanogaster stocksand crosses were reared at 25�, or at 18� where indicated, onyeast/glucose/wheat flour medium under standard condi-tions. Most stocks were obtained from the BloomingtonDrosophila Stock Center (Indiana University). The stocks

w ; P{sqhAA} and w ; P{sqhEE} were kindly provided by RogerKaress. The allele puc JK0 was generated by imprecise excision ofl(3)j4E1 with P{D2-3} transposase (Laski 1986). The full geno-type for flw 1 is y, cho, flw1. Wings from adult flies were dissectedin 70% ethanol and mounted in mounting medium (85%glycerol, 2.5% n-propylgallate).

X-gal staining of wing imaginal discs: Wing imaginal discsfrom third instar wandering larvae were dissected in PBS(137 mm NaCl, 3 mm KCl, 10 mm Na2HPO4, 2 mm KH2PO4),fixed 15 min in 4% paraformaldehyde in PBS, and washed withX-gal buffer (7.2 mm Na2HPO4, 2.8 mm NaH2HPO4, 1 mmMgCl2, 15 mm NaCl) before being incubated at 37� in X-galstaining solution [X-gal buffer, 5 mm K4Fe(CN)6, 5 mmK3Fe(CN)6, 2% X-gal]. The staining reaction lasted between1 and 4 hr. The tissues were washed in X-gal buffer andmounted in mounting medium.

SDS–polyacrylamide gel electrophoresis and immunoblot-ting: Twenty wing imaginal discs were taken up in 10 mm Tris–HCl, pH 6.8, 180 mm KCl, 50 mm NaF, 1 mm NaVO4, 10 mmb-glycerolphosphate, 1% Triton X-100, and 0.1% Tween 20and stored at�80�. An equal volume of 23 SDS sample buffer(100 mm Tris–HCl, pH 6.8, 200 mm DTT, 4% SDS, 20%glycerol, bromophenol blue) was added, and the proteins wereboiled for 5 min before loading on a 10% SDS–polyacrylamidegel. Separation of mono- and diphosphorylated JNK wasachieved by SDS–polyacrylamide gel electrophoresis (SDS–PAGE) at low voltage (50–60 V). Separated proteins weretransferred onto Immobilon-P PVDF membrane (Millipore,Bedford, MA). The membrane was blocked with 5% fat-freepowdered milk in TBST (137 mm NaCl, 3 mm KCl, 25 mm Tris,0.1% Tween 20) and incubated with the appropriate antibody[1:500 rabbit polyclonal a-P-JNK (Promega, Madison, WI),1:1000 rabbit polyclonal a-JNK (Santa Cruz Biotechnology),1:2000 a-tubulin (Sigma, St. Louis), and 1:10,000 horseradish-peroxidase-(HRP)-conjugated secondary antibody (Sigma)].HRP was detected with Supersignal West Pico (Pierce, Rock-ford, IL) and blue sensitive X-ray film (GRI).

RESULTS

puckered is a genetic enhancer of flapwing: flw 1 is aweak viable allele of Drosophila PP1b with indirectflight muscle defects (Raghavan 2000), but otherwisenormal morphology (Figure 1, A and B) and viability. Inthe process of analyzing mutants for the PP1c inhibitorInhibitor-2 at 67C10, we found that Df(3L)AC1 (67A02-67D13) genetically enhances flw 1 to complete lethality.Further analysis in the 67A–D region identified theEMS-induced mutation l(3)67BDn (Leicht 1988) to belethal over Df(3L)AC1 and to strongly enhance flw 1.l(3)67BDn/1 heterozygotes have normal viability andwings (not shown), whereas flw 1/Y ; l(3)67BDn/1 wassemilethal (90% of pupae fail to eclose), and survivingmales exhibited strong wing defects (Figure 1C) andoccasional third leg malformations (Figure 1C, arrow).Further analysis revealed that l(3)67BDn had a lethalmutation in the 67B–D region (below), but meiotic re-combination mapping showed that the flw enhancingmutation was located between the markers scarlet(73A03) and curled (86D03-04). Deficiency mapping be-tween those markers placed the flw enhancer in or nearto 84F02. By testing available mutants in this region, wefound that two mutants in puc at 84E10-11 (puc A251 and

1742 J. Kirchner et al.

puc H246), as well as l(3)j4E1, a P{lacW} insertion in thesecond intron of puc (Figure 1E; Spradling 1999), failedto complement l(3)67BDn (not shown). All these allelesshowed similar mutant phenotypes to each other andl(3)67BDn when combined with flw 1 (e.g., Figure 1D),although they were generally weaker than l(3)67BDn.flw1/ Y ; puc A251/1; for example, they had normal viability,and only �30% exhibited wing phenotypes that rangedfrom slight defects in the posterior part of the wing tostumps. The third leg malformation (Figure 1D, arrow)was visible only with the most severe wing phenotypes.We hypothesized that the stronger enhancement of flw 1

by l(3)67BDn was due to second-site mutations on thel(3)67BDn chromosome and extensively cleaned up thel(3)67BDn chromosome by recombination, exchangingthe 67A–67D region in the process. This reverted thelethality over Df(3L)AC1, suggesting that the originall(3)67BDn chromosome carries at least one additionalmutation, possibly in the 67A–67D region. Df(3L)AC1 wasviable over puc A251, which suggested that the Df(3L)AC1chromosome does not carry an allele of puc. The newl(3)67BDn recombinant genetically interacted with flw inthe same way as puc A251 and puc H246 (not shown). Weconcluded that the primary flw enhancing mutation onthe l(3)67BDn chromosome was an allele of puc andcalled it puc l(3)67BDn.

To test whether the genetic interaction between flw 1

and l(3)j4E1 was due to the P insertion in puc, weremobilized the P element and tested the recovered w�

excision chromosomes for genetic interaction withpuc A251 and flw 1. While l(3)j4E1 did not complementpuc A251, we found that the majority of excisions wereviable over puc A251 and did not enhance flw 1, provingthat l(3)j4E1 indeed carries a P-element-induced alleleof puc and that this is responsible for the genetic inter-action with flw. However, all of these excision derivativeswere lethal as homozygotes, which suggests that thel(3)j4E1 chromosome carries a second lethal mutation.One w� excision line, puc JK0, failed to complementpuc A251 and enhanced flw 1 in the same way as the otherpuc alleles. Molecular analysis identified a 2.2-kb de-letion with the left breakpoint in the second puc intronat the original P-element insertion site and the rightbreakpoint in the third exon, deleting the first 51% ofthe puc phosphatase catalytic domain (Figure 1E).

Since the amino acid sequences of all PP1c proteins areextremely similar (Dombrádi 1993), we tested whetherthe genetic interaction between flw and puc is specific toflw. puc A251 did not show any discernible mutant pheno-type with Pp1-87B87Bg-3, an allele of the major PP1c gene(not shown). Therefore, we concluded that puc specifi-cally enhances flw.

Expression of constitutively active Hemipterous(hepCA) enhances flw 1: puc codes for a dual specific-ity phosphatase that dephosphorylates DrosophilaJNK (bsk) (Martin-Blanco et al. 1998). We wonderedwhether other members of the Drosophila JNK pathway

also genetically interact with flw. The enzymatic antag-onist to the JNK phosphatase Puc is the JNK kinase Hep(Glise 1995). We expressed constitutively active Hep(UAS-hepCA) with the wing-specific driver vg-Gal4 in thedeveloping wing. At 25�, vg-Gal4/1 ; UAS-hepCA/1(vg.hepCA) was lethal, but at 18� it was viable and ex-hibited some wing defects, most notably missing poste-rior wing-vein material, notches in the wing margins,and held-out wings (Figure 2B). This phenotype wasenhanced in flw 1 hemizygous background. flw 1/Y ;vg.hepCA wings (Figure 2D) were much smaller thanthe wings of their flw 1/1 female siblings (Figure 2C)and were frequently notched and blistered. Further-more, flw 1/Y ; vg.hepCA were only 10% viable relative toflw 1/1 ; vg.hepCA (not shown). Thus, expression ofhepCA produced wing phenotypes that overlapped withflw 1/Y ; puc/1 in as much as posterior wing tissue waslost (see Figure 3D), and it also enhanced flw 1, eventhough the phenotypes of flw 1/Y ; vg.hepCA differedslightly from flw 1/Y ; puc/1.

basket (DJNK) mutants suppress flw6: Both puc mu-tants and hepCA overexpression should shift the equi-librium of nonphosphorylated (inactive) JNK andphosphorylated (active) JNK toward the latter. We

Figure 1.—(A–D) puc enhances flw. Adult males of (A)Oregon-R (OreR; wild type); (B) flw 1/ Y; (C) flw 1/ Y ; l(3)67BDn/1; and (D) flw 1/ Y ; puc A251/1. flw 1/ Y have wild-typewings (B), whereas introducing a mutant copy of eitherl(3)67BDn or pucA251 leads to severe wing defects (C and D)and occasional leg deformations (arrows). l(3)67BDn/1 andpuc A251/1 appear wild type (not shown). (E) puc gene struc-ture and mutants. Coding regions (black) and untranslatedregions (lavender) are indicated. l(3)j4E1 and puc A251 are P in-sertions in the second intron of puc. puc JK0 is a 2.2-kb deletion.

PP1b Regulates JNK 1743

therefore wondered whether hyperactivation of JNKwas the molecular cause of the genetic interaction withflw 1. If so, reducing the level of JNK should reduce theamount of active JNK and might thereby have a sup-pressing effect on strong flw mutants. We found that twoalleles of basket (bsk1 and bsk2) both suppressed thesemilethality of the strong allele flw6. The survival rateof flw6/Y is ,1%, whereas the survival rate of flw6/Y ;bsk1/1 was �26% (Table 1). bsk1 also suppressed thestrong and semilethal allele flw7 (not shown).

Regulation of nonmuscle myosin activity is the onlyessential, nonredundant function of flw in Drosophila(Vereshchagina 2004). The fact that bsk could sup-press flw6, albeit more weakly than, for example, the flw6

suppressors zip or P{sqhAA} (Vereshchagina 2004),could mean that nonmuscle myosin regulation andthe JNK pathway are biochemically linked. JNK couldwork either upstream of flw, for example, as an (in-direct) activator of nonmuscle myosin, or downstreamof flw, partly mediating a lethal response initiated byhyperactive Zip.

Flw is a negative regulator of puc expression in wingimaginal discs: The allele puc A251 is due to a P{lArB}insertion in the second intron of puc (Figure 1E) andhas also been widely used as a puc enhancer trap, as theexpression of the lacZ gene on the P element is underthe control of a genomic puc enhancer (Dobens 2001).X-gal staining showed that the only lacZ expression inwing imaginal discs of puc A251 heterozygotes was back-ground staining at the tip of the presumptive notum.However, in flw 1/Y ; puc A251/1 wing imaginal discs, lacZwas ectopically expressed (Figure 3, A and C), suggest-ing that flw acts as a negative regulator of puc expressionin wing imaginal discs. Since puc encodes a JNK phos-phatase, we hypothesized that this upregulation of pucexpression might be due to ectopic activation of JNK,which has been shown to induce apoptosis in wingmaginal discs (Adachi-Yamada et al. 1999a). In flw 1/Y ; puc A251/1 wing imaginal discs, lacZ was expressed in a

distinct region that will later develop into the posteriorpart of the adult wing; this was also the area that wasmost affected in the mutant wings (Figure 3D).

Flw negatively regulates puc expression through Bskand Zip: The ectopic expression of puc in flw 1/Y; puc A251/1

Figure 2.—Overexpression of constitutively active hemipter-ous in the wing (vg.hepCA) enhances flw 1. (A) OreR. (B)vg.hepCA. (C) flw 1/1 ; vg.hepCA. (D) flw 1/Y ; vg.hepCA. Wingsof flw 1/1 ; vg.hepCA exhibit blisters (arrow) and notchedwing margins and resemble those of vg.hepCA (B and C),whereas flw 1/ Y ; vg.hepCA wings are much smaller (D). B–D were reared at 18�, as vg.hepCA is lethal at 25�.

Figure 3.—puc A251 expression in wing imaginal discs andwing phenotypes. puc A251-lacZ expression in wing imaginaldiscs (A, C, E, G, I, and K) and adult wings (B, D, F, H, J,and L). (A and B) puc A251/TM6B. (C and D) flw 1/ Y ;puc A251/1. (E and F) flw 1/ Y ; bsk1/1 ; puc A251/1. (G andH) flw 1/ Y ; zip1/1 ; puc A251/1. (I and J) flw 1/ Y ; puc A251/P{sqhAA}. (K and L) puc A251/P{sqhEE}. Ectopic lacZ staining isvisible in flw 1/ Y ; puc A251/1 wing imaginal discs (C; compareto A), and the resulting wings show various defects (D, missingposterior wing tissue). bsk1 and zip1 each abolish the ectopiclacZ staining (E and G) and the resulting wings look normal(F and H). P{sqhAA} inhibits neither the ectopic lacZ stainingnor the wing phenotype of flw 1/ Y ; puc A251/1 (I and J).P{sqhEE} does not induce lacZ staining in a puc A251 mutant back-ground (K), and puc A251/P{sqhEE} wings appear wild type (L).

1744 J. Kirchner et al.

suggested that flw acts upstream of puc expression. Wehypothesized that flw might act upstream of JNK andtested this by introducing bsk1 into a flw 1/Y ; puc A251/1background. In flw 1/Y ; bsk1/1 ; puc A251/1 wing imagi-nal discs we no longer observed ectopic lacZ expression,and the adult wings of this genotype appeared wild type(Figure 3, E and F). This argued that flw regulated pucexpression through JNK in wing imaginal discs and thatJNK might become hyperactivated in flw 1/ Y ; puc A251/1flies.

We wondered in what way flw acts on JNK and puc ex-pression. PP1c dephosphorylates many different pro-teins, but, for the majority of these functions, we assumethat flw is redundant with the other PP1c forms,particularly Pp1-87B, which is present at much higherlevels (Dombrádi 1990). puc, however, genetically in-teracted specifically with flw and not with Pp1-87B. Sincethe single essential nonredundant function of Flw isinhibition of nonmuscle myosin activity, we wonderedwhether Flw might act on JNK and puc through non-muscle myosin regulation. To test this, we introducedthe mutant zip1 into flw 1/ Y ; puc A251/1 background andtested for lacZ expression. Most flw 1/ Y ; zip1/1 ; puc A251/1 wing imaginal discs did not show the ectopic expressionof lacZ characteristic of flw 1/ Y ; puc A251/1 (Figure 3G). Asmall number of wing imaginal discs showed ectopic lacZexpression, but the staining was weaker and the affectedarea was always much smaller than in flw 1/ Y ; puc A251/1wing imaginal discs (not shown). The adult wings of thisgenotype appeared wild type (Figure 3H). Altogether, thisshowed that flw regulated puc expression through zip aswell as bsk and that zip, like flw, acts upstream of theregulation of puc expression.

Mbs3 enhances flw 1/ Y ; puc A251/1: As described above,flw regulates puc expression through zip in wing imaginaldiscs. We wondered whether zip1 suppressed flw 1/ Y ;puc A251/1 by compensating for a moderate hyperactiva-

tion of Zip in this mutant background. If so, we spe-culated that further activation of Zip in flw 1/ Y ; puc A251/1might enhance the phenotype. Disrupting the functionof the nonmuscle myosin phosphatase by mutations inthe myosin phosphatase-targeting subunit should lead toincreased activation of Zip. Unfortunately, no mutantshave been described for the Flw-specific MYPT-75D.However, we found that the genotype flw 1/ Y ; 1, puc A251/Mbs 3, 1 is lethal (Table 2), providing further genetic evi-dence that hyperactivation of nonmuscle myosin canactivate JNK.

Sqh phosphorylation mutants do not alter puc ex-pression: Mutations in strong flw alleles lead to hyper-phosphorylation of Sqh on T21 and S22, which induceshyperactivation of Zipper (Vereshchagina 2004). Con-structs expressing Sqh T21 and S22 phosphorylationmutants genetically interact with flw. In P{sqhAA}, T21and S22 are mutated to alanine; therefore SqhAA cannotbe activated, while in P{sqhEE} the same residues aremutated to glutamic acid, which mimics phosphoryla-tion ( Jordan 1997). P{sqhAA} suppresses the semilethal-ity of flw6, whereas P{sqhEE} enhances flw 1 to completelethality, suggesting that P{sqhAA} and P{sqhEE} have adominant-negative and constitutively active effect, re-spectively, although neither show a phenotype in a wild-type background. We used P{sqhAA} and P{sqhEE} to testwhether hyperphosphorylation of Sqh could be themechanism by which puc expression becomes elevatedin flw 1/ Y ; puc A251/1. Both transgenes were expressedfrom their native promoter in a sqh1 background. If theectopic expression of puc in flw 1/ Y ; puc A251/1 wingimaginal discs was due to hyperphosphorylation ofSqh, introducing P{sqhAA} into this background mightabolish the expression of puc. Conversely, expression

TABLE 1

bsk1 suppresses flw6 (from cross flw6/FM7c 3 bsk1/CyO,act-GFP)

F1 genotype No. of adults

flw6/Y ; bsk1/1 14flw6/Y ; CyO, act-GFP/1 1FM7c/Y ; bsk1/1 26FM7c/Y ; CyO, act-GFP/1 31flw6/1 ; bsk1/1 53flw6/1 ; CyO, act-GFP/1 45FM7c/1 ; bsk1/1 37FM7c/1 ; CyO, act-GFP/1 60Total 267

flw6 is a semilethal mutant with a survival rate of ,1%. In-troducing the allele bsk1 into a flw6 mutant background sup-presses the semilethality of flw6 (compare the two underlinedgenotypes). flw6/ Y ; bsk1/1 show a viability of �26% (relativeto their female siblings, flw6/1 ; bsk1/1).

TABLE 2

Mbs 3 enhances flw 1/ Y ; puc A251/1 (from cross flw 1/FM7 ;puc A251/TM6B 3 Mbs3/TM6B,Ub-GFP)

F1 genotype No. of adults

flw1/ Y ; 1, puc A251/Mbs3, 1 0flw 1/ Y ; puc A251/TM6B, Ub-GFP 14flw 1/ Y ; Mbs3/TM6B 23FM7/ Y ; 1, puc A251/Mbs3, 1 29FM7/ Y ; puc A251/TM6B, Ub-GFP 13FM7/ Y ; Mbs3/TM6B 11flw 1/1 ; 1, puc A251/Mbs3, 1 50flw 1/1 ; puc A251/TM6B, Ub-GFP 26flw 1/1 ; Mbs3/TM6B 35FM7/1 ; 1, puc A251/Mbs3, 1 33FM7/1 ; puc A251/TM6B, Ub-GFP 24FM7/1 ; Mbs3/TM6B 19Total 277

Mbs3 enhances flw 1/ Y ; puc A251/1 to complete lethality (keygenotype is underlined), whereas flw 1/ Y ; Mbs3/TM6B andflw 1/1 ; 1, puc A251/Mbs 3, 1 show normal viability (as shown)and phenotype (not shown). This suggests that flw 1 enhancespuc through nonmuscle myosin.

PP1b Regulates JNK 1745

of P{sqhEE} might induce ectopic lacZ expression in apuc A251 background. We found that neither was the case;there was still ectopic lacZ staining in flw 1/ Y ; puc A251/P{sqhAA} wing imaginal discs (Figure 3I), and the re-sulting wings looked similar to flw 1/ Y ; puc A251/1(Figure 3J). No ectopic lacZ expression was visible inpuc A251/P{sqhEE} wing imaginal discs, and the wings ofthis genotype resembled wild type (Figure 3, K and L).Together, this suggested that hyperphosphorylation ofSqh at T21 and S22 did not cause ectopic puc expressionin flw 1/ Y ; puc A251/1 wing imaginal discs.

The level of phospho-JNK is elevated in a flw 1/ Y ;puc A251/1 mutant background: As shown in Figure 3E,introducing bsk1 into a flw 1/ Y ; puc A251/1 backgroundabolished ectopic puc-lacZ staining in wing imaginaldiscs. On a molecular level, this could mean that JNKis ectopically activated in flw 1/ Y ; puc A251/1 flies andactivates puc-lacZ expression. JNK is activated by phos-phorylation on T181 and Y183, which are located in aconserved loop of the kinase subdomain VIII (corre-sponding to T183 and Y185 in mammalian JNK1)(Derijard 1994). With an anti-P-JNK antibody raisedagainst dually phosphorylated mammalian JNK thatcross-reacts with Drosophila phosphorylated (phos-pho-) JNK (Tateno 2000), we examined phospho-JNKlevels in extracts from wing imaginal disc. As a control,we used MS1096-Gal4/1 ; UAS-hepCA/1 (MS1096.hepCA),which drives constitutively active Hep in wing imaginaldiscs, resulting in high levels of phospho-JNK (Figure4A). In flw 1/ Y ; puc A251/1 extracts, there were elevatedlevels of phospho-JNK, but not in flw 1/1 ; puc A251/1and wild type. We also noted the presence of a band ofslightly faster mobility in MS1096.hepCA, flw 1/ Y ;puc A251/1, flw 1, and puc A251/TM6B, but not in wild type(Figure 4A). We assume that this band corresponds tomono-phosphorylated JNK, to which the antiphospho-JNK antibody has a low cross-reactivity. Only duallyphosphorylated JNK is active (Anderson 1990). Takentogether, JNK is ectopically activated in flw 1/ Y ; puc/1wing imaginal discs, but not in flw 1 and puc A251/TM6B.

DISCUSSION

In this study, we show that the nonmuscle myosinphosphatase flw interacts genetically with componentsof the JNK signaling pathway. The proteins JNK andPP1b (Flw), as well as the mechanisms of JNK sig-nal transduction and nonmuscle myosin activation, arehighly conserved between Drosophila and humans. Thissuggests that our findings of Drosophila PP1b regulatingJNK through nonmuscle myosin may be relevant forsimilar processes in human cells and tissues.

bsk ( JNK) mutants suppressed the semilethality of thestrong allele flw6, while puc and constitutively active hepCA

enhanced the weak viable allele flw 1, resulting in severewing defects. The level of diphosphorylated (active)

JNK was elevated in a flw 1/ Y ; puc A251 mutant back-ground. This, together with the finding that puc expres-sion was upregulated in wing imaginal discs of flw 1/ Y ;puc A251/1, implies that flw can act as a negative reg-ulator of JNK. This was further supported by the fact thatpuc expression was not upregulated in flw 1/ Y ; bsk1/1 ;puc A251/1 wing imaginal discs, where reduced amountsof overall JNK (Bsk) protein probably compensate forelevated activity of JNK in flw 1/ Y ; puc A251/1. A possibledifficulty with using puc A251 [or puc E69, another fre-quently used puc enhancer trap (Martin-Blanco et al.1998)] as a reporter of puc expression is that both linesare also puc mutants, and puc probably regulates its ownexpression through a negative feedback loop involvingJNK (Figure 4B). However, we confirmed in an in-dependent assay with an anti-P-JNK antibody that JNKwas indeed ectopically activated in flw 1/ Y ; puc A251/1.

Figure 4.—(A) JNK is ectopically activated in flw 1/Y ;puc A251/1 wing imaginal discs. antiphospho-JNK detects du-ally phosphorylated (active) JNK in the control MS1096.hep-CA and flw 1/Y ; puc A251/1, but not in flw 1/1 ; puc A251/1, flw 1/Y, puc A251/TM6B, and wild type (OreR). A band of slightly fast-er mobility, probably corresponding to monophosphorylatedJNK, can be detected in MS1096.hepCA, flw 1/Y ; puc A251/1,flw 1/Y, and puc A251/TM6B. (B and C) Model of interaction be-tween Flw and JNK in wild type (B) and flw 1/Y ; puc/1 (C).Active signaling is indicated by solid arrows, nonactive signal-ing by shaded arrows. Zip, Flw, and Mbs presumably act up-stream of Hep. (B) Inhibition of Zip through Flw and Mbs,as well as wild-type puc, prevent aberrant activation of JNK(Bsk). (C) Failure of Zip inhibition through flw 1, and ofJNK dephosphorylation through a puc mutation, lead to ele-vated levels of phosphorylated and activated JNK in flw 1/Y ;puc/1, inducing high expression of puc. di-P-JNK, diphos-phorylated JNK; mono-P-JNK, monophosphorylated JNK.

1746 J. Kirchner et al.

Wing imaginal disc extracts from both flw 1 and puc A251/1showed somewhat elevated levels of monophosphory-lated, but not diphosphorylated, JNK (Figure 4A). Inflw 1/ Y ; puc A251/1, we suggest that dephosphorylationof JNK fails to such an extent that JNK is substantiallydiphosphorylated and thereby activated (Figure 4C).Since it has been shown that activation of JNK in wingimaginal discs induces apoptosis (Adachi-Yamada et al.1999a,b), it is likely that the flw 1/ Y ; puc/1 wing phe-notype is due to increased death of cells with aberrantlyactivated JNK.

A zip mutant suppressed the upregulation of puc inwing imaginal discs as well as the adult wing phenotypeof flw 1/ Y ; puc A251/1. This shows that nonmuscle myosinacts upstream of JNK and mediates an activating signalon JNK in a flw 1/ Y ; puc/1 mutant background (Figure4B); this is also consistent with our finding that a mutantin the myosin phosphatase-targeting subunit Mbs en-hanced flw 1/ Y ; puc A251/1 to lethality. Drosophila en-codes two myosin phosphatase-targeting subunits, Mbsand MYPT-75D. Mbs binds both Flw and Pp1-87B, whileMYPT-75D binds Flw specifically (Vereshchagina 2004).Unfortunately, we were not able to test for genetic inter-action between flw 1/ Y ; puc A251/1 and MYPT-75D becauseno MYPT-75D mutants have been described. We alsofound that a Rho1 mutant abolished ectopic lacZ stainingin flw 1/Y ; puc A251 wing imaginal discs (not shown). Rho1 isan activator of Rho-dependent kinase, which phosphor-ylates and activates myosin light chain, as well as phos-phorylating and inhibiting Mbs (Derijard 1994; Mizuno1999). Furthermore, both zip and Rho1, in combinationwith other genetic interactors, have been reported toshow a malformed third-leg phenotype that resemblesthat of flw 1/ Y ; puc/1 (Halsell 1998; Bayer 2003).

Dorsal closure and wound healing depend on bothnonmuscle myosin and JNK activity, but a clear geneticor molecular interaction between these pathways hasnot been previously demonstrated. This is the first studythat shows that disruption of nonmuscle myosin caninduce activation of JNK, although the mechanism ofsignal transduction is not clear. The single essential andnonredundant function of flw is the inhibition of non-muscle myosin activity, presumably by dephosphorylat-ing Sqh at T21 and S22. However, our results suggestthat hyperphosphorylation of Sqh at these residues maynot explain the elevated expression of puc in flw 1/ Y ;puc A251/1 flies. Additional mechanisms may exist toregulate nonmuscle myosin assembly and activitythrough phosphorylation (e.g., Nishikawa 1984; Ikebe1990; Murakami 1998), and the complex Flw/Mbs mayhave other targets in the actomyosin network in addi-tion to Sqh. For example, moesin and focal adhesionkinase are potential targets for mammalian myosinphosphatase (Fukata 1998; Fresu 2001). Interestingly,overexpression of the focal adhesion protein tensin(blistery) in Drosophila wing imaginal discs activates JNKand induces apoptosis (Lee et al. 2003).

Two important questions remain unanswered re-garding our results on the activation of JNK throughnonmuscle myosin. First, what is the molecular path-way from nonmuscle myosin to JNK? Several mecha-nisms have been identified that activate JNK andinduce apoptosis in wing imaginal discs. For example,mutations in the caspase inhibitor DTraf1 (Drosophilatumor necrosis factor receptor-associated factor 1), as wellas inhibition of DTraf1 by overexpression of Hid(head involution defective), Rpr (reaper), or Grim, inducesJNK-mediated apoptosis (Kuranaga 2002; Ryoo2004), possibly through Msn (misshapen) or Ask1(apoptosis signal-regulating kinase 1) (Liu 1999; Kuranaga2002). Other factors involved in inducing apoptosisand activating JNK are Eiger (Drosophila tumor-necrosisfactor superfamily ligand) (Igaki 2002), its receptorWengen (Kauppila 2003), Src42A (Tateno 2000), theserine C-palmitoyltransferase Lace (Adachi-Yamadaet al. 1999b), Blistery (Drosophila Tensin) (Lee et al.2003), and Decapentaplegic and Wingless (Adachi-Yamada et al. 1999a). It is not clear how these factorssignal to JNK or whether they act in a single pathway,let alone whether any of them interact with nonmusclemyosin. It is likely that there are several independentways of activating JNK, and it has been suggested thatinduction of apoptosis through JNK activation is aregulatory mechanism to eliminate abnormally devel-oping cells in wing imaginal discs (reviewed in Adachi-Yamada and O’Connor 2004). This leads directly to oursecond question: Are our results of nonmuscle myosinsignaling to JNK significant in a developmental context?The fact that bsk1 suppressed the semilethality of flw6

indicates that the interactions that we have uncoveredare not confined to our main experimental model ofectopic JNK activation in wing imaginal discs. Theobvious system for studying possible interactions be-tween nonmuscle myosin and JNK would be dorsalclosure, which depends on both the coordinatedassembly and the contraction of the actomyosin cyto-skeleton and on activation of JNK (reviewed in Jacinto2002). So far, there is no evidence that dorsal closure isaffected in flw mutant embryos; however, there is amaternal contribution of flw that may conceal embry-onic phenotypes. Actomyosin and JNK do not promotedorsal closure completely independently from eachother; for example, the expression of many componentsof the actomyosin cytoskeleton is upregulated in re-sponse to JNK (Jasper 2001). Also, actin and myosinfailed to accumulate along the leading edge of theepidermis in the puc mutant background (Martin-Blanco et al. 1998). Both findings would place theactomyosin cytoskeleton downstream of JNK, whereasour genetic data place flw and zip upstream of JNK andpuc. It is possible, however, that actomyosin acts bothupstream and downstream of JNK during dorsal closure.Because of the conserved nature of the componentsinvolved, it is likely that our finding that nonmuscle

PP1b Regulates JNK 1747

myosin can signal to and activate JNK is relevant tofurthering our understanding of processes like dorsalclosure and wound healing in Drosophila and humans.

The authors thank Karen Clifton for technical assistance, RogerKaress for providing P{sqhAA} and P{sqhEE}, and the Bloomington StockCenter for providing the mutant fly strains. We also thank HelenWhite-Cooper and Esther Duperchy for help and advice and criticalreading of this manuscript. This work was supported by grants fromthe United Kingdom Biotechnology and Biological Sciences ResearchCouncil and the United Kingdom Medical Research Council. D.B. is aTodd Bird Research Fellow at New College, Oxford.

LITERATURE CITED

Adachi-Yamada, T., and M. B. O’Connor, 2004 Mechanisms for re-moval of developmentally abnormal cells: cell competition andmorphogenetic apoptosis. J. Biochem. 136: 13–17.

Adachi-Yamada, T., K. Fujimura-Kamada, Y. Nishida and K.Matsumoto, 1999a Distortion of proximodistal informationcauses JNK-dependent apoptosis in Drosophila wing. Nature400: 166–169.

Adachi-Yamada, T., T. Gotoh, I. Sugimura, M. Tateno, Y. Nishidaet al., 1999b De novo synthesis of sphingolipids is requiredfor cell survival by down-regulating c-Jun N-terminal kinase inDrosophila imaginal discs. Mol. Cell. Biol. 19: 7276–7286.

Anderson, N. G., J. L. Maller, N. K. Tonks and T. W. Sturgill,1990 Requirement for integration of signals from two distinctphosphorylation pathways for activation of MAP kinase. Nature343: 651–653.

Bayer, C. A., S. R. Halsell, J. W. Fristrom, D. P. Kiehart and L.von Kalm, 2003 Genetic interactions between the RhoA andStubble-stubbloid loci suggest a role for a type II transmembraneserine protease in intracellular signaling during Drosophila ima-ginal disc morphogenesis. Genetics 165: 1417–1432.

Bollen, M., 2001 Combinatorial control of protein phosphatase-1.Trends Biochem. Sci. 26: 426–431.

Cohen, P. T. W., 2002 Protein phosphatase 1—targeted in many di-rections. J. Cell Sci. 115: 241–256.

Derijard, B., M. Hibi, I. H. Wu, T. Barrett, B. Su et al., 1994 JNK1:a protein kinase stimulated by UV light and Ha-Ras that bindsand phosphorylates the c-Jun activation domain. Cell 76:1025–1037.

Dobens, L. L., E. Martin-Blanco, A. Martinez-Arias, F. C. Kafatosand L. A. Raftery, 2001 Drosophila puckered regulates Fos/Jun levels during follicle cell morphogenesis. Development128: 1845–1856.

Dombrádi, V., J. M. Axton, H. M. Barker and P. T. Cohen,1990 Protein phosphatase 1 activity in Drosophila mutants withabnormalities in mitosis and chromosome condensation. FEBSLett. 275: 39–43.

Dombrádi, V., D. J. Mann, R. D. C. Saunders and P. T. W. Cohen,1993 Cloning of the fourth functional gene for protein phos-phatase 1 in Drosophila melanogaster from its chromosomal loca-tion. Eur. J. Biochem. 212: 177–183.

Fresu, M., M. Bianchi, J. T. Parsons and E. Villa-Moruzzi,2001 Cell-cycle-dependent association of protein phosphatase1 and focal adhesion kinase. Biochem J. 358: 407–414.

Fukata, Y., K. Kimura, N. Oshiro, H. Saya, Y. Matsuura et al.,1998 Association of the myosin-binding subunit of myosinphosphatase and moesin: dual regulation of moesin phosphory-lation by Rho-associated kinase and myosin phosphatase. J. CellBiol. 141: 409–418.

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

Halsell, S. R., and D.P. Kiehart, 1998 Second-site noncomple-mentation identifies genomic regions required for Drosophilanonmuscle myosin function during morphogenesis. Genetics148: 1845–1863.

Igaki, T., H. Kanda, Y. Yamamoto, H. Kanuka, E. Kuranagaet al., 2002 Eiger, a TNF superfamily ligand that triggers theDrosophila JNK pathway. EMBO J. 21: 3009–3018.

Ikebe, M., and S. Reardon, 1990 Phosphorylation of bovine plateletmyosin by protein kinase C. Biochemistry 29: 2713–2720.

Jacinto, A., S. Woolner and P. Martin, 2002 Dynamic analysis ofdorsal closure in Drosophila: from genetics to cell biology. Dev.Cell 3: 9–19.

Jasper, H., V. Benes, C. Schwager, S. Sauer, S. Clauder-Munsteret al., 2001 The genomic response of the Drosophila embryo toJNK signaling. Dev. Cell 4: 579–586.

Jordan, P., and R. Karess, 1997 Myosin light chain-activating phos-phorylation sites are required for oogenesis in Drosophila. J. CellBiol. 139: 1805–1819.

Jordan, P., S. Wheatley and R. Karess, 1995 Analysis of spaghetti-squash reveals a role for nonmuscle myosin II in cytokinesis,embryogenesis, and oogenesis, 36th Annual Drosophila ResearchConference, Atlanta, Vol. 36, p. 187A.

Kauppila, S., W. S. Maaty, P. Chen, R. S. Tomar, M. T. Eby et al.,2003 Eiger and its receptor, Wengen, comprise a TNF-like sys-tem in Drosophila. Oncogene 22: 4860–4867.

Kockel, L., J. G. Homsey and D. Bohmann, 2001 Drosophila Ap-1:lessons from an invertebrate. Oncogene 20: 2347–2364.

Kuranaga, E., H. Kanuka, T. Igaki, K. Sawamoto, H. Ichijo et al.,2002 Reaper-mediated inhibition of DIAP1-induced DTRAF1degradation results in activation of JNK in Drosophila. Nat. CellBiol. 4: 705–710.

Laski, F. A., D. C. Rio and G. M. Rubin, 1986 Tissue specificity ofDrosophila P element transposition is regulated at the level ofmRNA splicing. Cell 44: 7–19.

Lee, A., and J. E. Treisman, 2004 Excessive myosin activity in mbsmutants cause photoreceptor movement out of the Drosophilaeye disc epithelium. Mol. Biol. Cell 15: 3285–3295.

Lee, S. B., K. S. Cho, E. Kim and J. Chung, 2003 blistery encodesDrosophila tensin protein and interacts with integrin and theJNK signaling pathway during wing development. Development130: 4001–4010.

Leicht, B. G., and J. J. Bonner, 1988 Genetic analysis of chromosomalregion 67A-D of Drosophila melanogaster. Genetics 119: 579–593.

Lin, Q., E. S. Buckler, S. V. Muse and J. C. Walker, 1999 Molecularevolution of type 1 serine/threonine protein phosphatases. Mol.Phylogenet. Evol. 12: 57–66.

Liu, H., Y. Su, E. Becker, J. Treisman and E.Y. Skolnik, 1999 ADrosophila TNF-receptor-associated factor (TRAF) binds theSte20 kinase Misshapen and activates Jun kinase. Curr. Biol. 9:101–104.

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

Mizuno, T., M. Amano, K. Kaibuchi and Y. Nishida, 1999 Iden-tification and characterization of Drosophila homolog of Rho-kinase. Gene 238: 437–444.

Mizuno, T., K. Tsutsui and Y Nishida, 2002 Drosophila myosinphosphatase and its role in dorsal closure. Development 129:1215–1223.

Murakami, N., V. P. Chauhan and M. Elzinga, 1998 Two non-muscle myosin II heavy chain isoforms expressed in rabbit brains:filament forming properties, the effects of phosphorylation byprotein kinase C and casein kinase II, and location of the phos-phorylation sites. Biochemistry 37: 1989–2003.

Nishikawa, M., J. R. Seller, R. S. Adelstein and H. Hidaka,1984 Protein kinase C modulates in vitro phosphorylation ofthe smooth muscle heavy meromyosin by myosin light chainkinase. J. Biol. Chem. 259: 8808–8814.

Raghavan, S., I. Williams, H. Aslam, D. Thomas, B. Szoor et al.,2000 Protein phosphatase 1beta is required for the mainte-nance of muscle attachments. Curr. Biol. 10: 269–272.

Ramet, M., Z. D. Lanot and P. Manfruelli, 2002 JNK signalingpathway is required for efficient wound healing in Drosophila.Dev. Biol. 241: 145–156.

Ryoo, H. D., T. Gorenc and H. Steller, 2004 Apoptotic cells caninduce compensatory cell proliferation through the JNK and theWingless pathways. Dev. Cell 7: 491–501.

Spradling, A. C., D. Stern, A. Beaton, E. J. Rhem, T. Laverty et al.,1999 The Berkeley Drosophila Genome Project gene disrup-tion project: single P-element insertions mutating 25% of vitalDrosophila genes. Genetics 153: 135–177.

1748 J. Kirchner et al.

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

Tateno, M., Y. Nishida and T. Adachi-Yamada, 2000 Regulationof JNK by Src during Drosophila development. Science 287:324–327.

Vereshchagina, N., D. Bennett, B. Szoor, J. Kirchner, S. Grosset al., 2004 The essential role of PP1beta in Drosophila is to reg-ulate non-muscle myosin. Mol. Biol. Cell 15: 4395–4405.

Yong, J., I. Tan, L. Lim and T. Leung, 2006 Phosphorylation of my-osin phosphatase targeting subunit 3 (MYPT3) and regulation of

protein phosphatase 1 by protein kinase A. J. Biol. Chem. 281:31202–31211.

Young, P. E., A. M. Richman, A. S. Ketchum and D. P. Kiehart,1993 Morphogenesis in Drosophila requires nonmuscle myosinheavy chain function. Genes Dev. 7: 29–41.

Zeitlinger, J., and D. Bohmann, 1999 Thorax closure in Drosoph-ila: involvement of Fos and the JNK pathway. Development 126:3947–3956.

Communicating editor: T. Schüpbach

PP1b Regulates JNK 1749