CORRECTION

Correction: Distinct roles and requirements for Ras pathwaysignaling in visceral versus somatic muscle founder specification(doi: 10.1242/dev.169003)Yiyun Zhou, Sarah E. Popadowski, Emily Deutschman and Marc S. Halfon

There was an error published in Development (2019). 146, dev169003 (doi: 10.1242/dev.169003).

The surname of author Emily Deutschman was spelled incorrectly (Emily Deustchman).

The correct name appears above and both the online full-text and PDF versions have been updated.

The authors apologise to readers for this mistake.

1

© 2019. Published by The Company of Biologists Ltd | Development (2019) 146, dev176743. doi:10.1242/dev.176743

DEVELO

PM

ENT

RESEARCH ARTICLE

Distinct roles and requirements for Ras pathway signaling invisceral versus somatic muscle founder specificationYiyun Zhou1, Sarah E. Popadowski1, Emily Deutschman1 and Marc S. Halfon1,2,3,4,5,*

ABSTRACTPleiotropic signaling pathways must somehow engender specificcellular responses. In the Drosophila mesoderm, Ras pathwaysignaling specifies muscle founder cells from among the broaderpopulation of myoblasts. For somatic muscles, this is an inductiveprocess mediated by the ETS-domain downstream Ras effectorsPointed and Aop (Yan). We demonstrate here that for the circularvisceralmuscles, despite superficial similarities, a significantly differentspecification mechanism is at work. Not only is visceral founder cellspecification not dependent on Pointed or Aop, but Ras pathwaysignaling in its entirety can be bypassed. Our results show that de-repression, not activation, is the predominant role of Ras signaling inthe visceral mesoderm and that, accordingly, Ras signaling is notrequired in the absence of repression. The key repressor actsdownstream of the transcription factor Lame duck and is likely amember of the ETS transcription factor family. Our findings fit with agrowing body of data that point to a complex interplay between the Raspathway, ETS transcription factors, and enhancer binding as a crucialmechanism for determining unique responses to Ras signaling.

KEY WORDS: Myogenesis, Signal integration, Founder cell,Muscle progenitor, Fusion competent myoblast

INTRODUCTIONEmbryonic development requires that individual cells within a fieldacquire new, distinct fates. The Drosophila larval musculature hasemerged as an exemplary system for investigating this process,revealing important insights into the acquisition of developmentalcompetence, progressive determination of cell fate, and theintegration of multiple signals at the transcriptional level. It hasproven to be a particularly effective model for understanding howspecific outcomes can be obtained from the activation of thereceptor tyrosine kinase (RTK)/Ras/mitogen-activated proteinkinase (MAPK) signaling cascade (Halfon et al., 2000), a highlypleiotropic pathway involved in numerous developmental processesand dysregulated in a wide set of developmental disorders andcancers (Imperial et al., 2017; Schlessinger, 2000; Tidyman andRauen, 2009).

In the somatic (body wall) musculature, which has been studiedmost extensively, individual syncytial muscle fibers develop via thefusion of two cell types drawn from an initially undifferentiated poolof myoblasts within the stage 10-11 (mid-embryogenesis)mesoderm: a single ‘founder cell’ (FC; itself the product of theasymmetric division of a muscle ‘progenitor’) and one or more‘fusion competent myoblasts’ (FCMs; Fig. 1) (for a review, seeDobi et al., 2015). Whereas FCMs are uncommitted, FCs areinduced with specific identities. FCMs fuse with FCs, with theresulting syncytium maintaining the identity provided by the FC.FC specification is thus a crucial step in muscle development as theFC genetic program controls muscle size, orientation, expression ofcell-surface proteins, etc.

Although multiple signaling pathways, including the Wg (Wnt)and Dpp (BMP) pathways, are integrated to specify particularmuscle fates, the key event in muscle determination is MAPKactivation via RTK/Ras pathway signaling (Buff et al., 1998;Carmena et al., 2002, 1998; Halfon et al., 2000). In the somaticmesoderm, the relevant receptors are the Drosophila EGF and FGFreceptor homologs. Cells that are competent to respond to Ras/MAPK signaling are induced as an equivalence group andsubsequently restricted by lateral inhibition (mediated by Notch-Delta signaling) to a single muscle progenitor.

These events have been studied in detail at the molecular level inthe context of the muscle-identity gene even skipped (eve). A 300 bptranscriptional enhancer directly integrates the inductive Ras/MAPK signaling with a combination of additional signal-derivedand tissue-specific transcription factors (TFs) to activate eveexpression (Halfon et al., 2000). The Ras/MAPK effector is theETS-domain TF Pointed (Pnt), which binds the enhancer at up toeight distinct sites (Boisclair Lachance et al., 2018; Halfon et al.,2000). In the absence of induction these sites are bound by the ETS-domain repressor Anterior open (Aop, also known as Yan) (Halfonet al., 2000; Webber et al., 2013; Boisclair Lachance et al., 2018).Recent evidence suggests that Pnt bound at these or other sites mayalso contribute to repression in the absence of MAPK activation(Webber et al., 2018). Importantly, experiments have shown thatinduction trumps repression: in the absence of both Pnt and Aopbinding, there is no gene activation (Halfon et al., 2000;unpublished data). Ectopic activation of the Ras/MAPK pathwayleads to ectopic FC formation in all competent cells, at the expenseof FCMs; this has been demonstrated at the level of the receptortyrosine kinases (activated EGFR and FGFR), of Ras (activatedRas) and of the effector (activated Pnt) (Carmena et al., 1998;Halfon et al., 2000).

We focus here on the circular visceral muscle fibers, whichsurround the midgut and develop from the trunk visceral mesoderm(for simplicity, wewill refer to these simply as ‘visceral muscle’ and‘visceral mesoderm’, respectively). These muscle fibers appear todevelop similarly to the somatic muscles (Fig. 1), with the exceptionthat they are only binucleate and it is unclear whether there is aReceived 14 June 2018; Accepted 19 December 2018

1Department of Biochemistry, University at Buffalo-State University of New York,Buffalo, NY 14203, USA. 2Department of Biological Sciences, University at Buffalo-State University of New York, Buffalo, NY 14203, USA. 3Department of BiomedicalInformatics, University at Buffalo-State University of New York, Buffalo, NY 14203,USA. 4NY State Center of Excellence in Bioinformatics and Life Sciences, Buffalo,NY 14203, USA. 5Molecular and Cellular Biology Department and Program inCancer Genetics, Roswell Park Comprehensive Cancer Center, Buffalo, NY 14263,USA.

*Author for correspondence (mshalfon@buffalo.edu)

M.S.H., 0000-0002-4149-2705

1

© 2019. Published by The Company of Biologists Ltd | Development (2019) 146, dev169003. doi:10.1242/dev.169003

DEVELO

PM

ENT

‘muscle progenitor’ cell specified prior to visceral FC specification(Martin et al., 2001). As with somatic FCs, visceral FC specificationoccurs following MAPK activation – here via the single signalingpair of the Anaplastic lymphoma kinase (Alk) receptor tyrosinekinase and its ligand Jelly belly (Jeb) – and, just as for the somaticmusculature, ectopic activation of the Ras/MAPK pathway causespresumptive FCMs to be re-specified as FCs (Fig. 2D) (Englundet al., 2003; Lee et al., 2003; Weiss et al., 2001). However, thedetails of these events have not been established.We now show that, despite the apparent similarities between

somatic and visceral FC specification, fundamental differences existwith respect to the role of Ras/MAPK signaling in specifying the FCfate. Unlike the positive inductive role for MAPK activity in thesomatic mesoderm, in the visceral mesoderm MAPK activity isinstead required to relieve repression of FC fates, and thetranscriptional effectors Pnt and Aop do not appear to play asignificant role in this process. Moreover, MAPK activity can bedispensed with entirely in the absence of the FCM-specific TFLame duck (Lmd) or when repressor-binding sites are mutated in anFC-specific enhancer for the mind bomb 2 (mib2) gene. Thus,unlike in the somatic mesoderm, Ras/MAPK signaling is notessential for visceral FC specification. Our results illustrate howsimilar-appearing developmental processes can result from differentunderlying molecular mechanisms and provide fresh insights intohow unique responses to Ras pathway signaling are determinedwithin similar cellular and developmental contexts.

RESULTSVisceral founder cell specification is independent of theETS-domain TFs Pnt and AopIn a previous screen for genes that respond differentially to differentRas pathway components, we observed that, despite responding toRTK and Ras activation, the FC gene mib2 is not regulated by theRas effector Pnt in the visceral mesoderm (Leatherbarrow and

Halfon, 2009). Expression of both mib2 RNA and a mib2-lacZreporter gene driven by an FC-specific enhancer (mib2-FCenhancer) is normal in pnt null mutant embryos (Fig. 2A,C,E;Halfon et al., 2011), and expression of a constitutively active form ofPnt (Pntact) has no effect on expression of either the endogenousgene or the reporter (Fig. 2F, Fig. S1D; Leatherbarrow and Halfon,2009). Similarly, mib2 expression in the visceral mesoderm isnormal in embryos mutant for the ETS-domain repressor aop (yan)(Fig. 2G, Fig. S1E), and in response to expression of theconstitutively repressing version yanact (Fig. 2H, Fig. S1F; Halfonet al., 2011). This raised the question of whether this is a mib2-specific regulatory effect, or whether these two Ras effectors, whichboth play a significant role in somatic FC determination, are notrequired for visceral FC specification.

To test this, we assessed the expression of additional visceral FCand FCM markers in some or all of the pnt null, aop null, pntact andyanact backgrounds. Expression of the somatic muscle-identity geneeven skipped (eve) was used as a control (insets in Fig. 2), as itsexpression is reduced or expanded in response to pnt and aop lossand gain of function in a well-defined way (Halfon et al., 2000).Expression of the FC markers org-1, kirre [also known as

Fig. 1. Overview of Drosophila muscle development. In both the somaticmesoderm (top) and trunk visceral mesoderm (bottom), initially equivalentmyoblasts (left panel) are fated to become either muscle founder cells (FCs;middle panel, gray, red and blue) or fusion competent myoblasts (FCMs;middle panel, yellow). FCs have specific identities, represented here bydifferent colors, conferred by the activity of ‘identity genes’ active in the FCs.FCMs fuse with FCs to generate individual muscle fibers (right panel), witheach fiber maintaining the fate provided by the founder cell. Fig. 2. mib2 expression responds to Ras signaling but not to pnt or aop.

All panels show stage 11 embryos stained for expression of Mib2 (using themib2_FCenhancer lacZ reporter, green) and the pan-visceral-mesodermmarker Biniou (Bin, magenta), except B and E, which show onlymib2 RNA bymeans of in situ hybridization. Insets show the somatic mesoderm expressionof Even skipped in a single dorsal cell cluster used as a control for the variousgenotypes (Halfon et al., 2000). (A) Wild-type (WT) embryo depicted ventralside up and anterior to the left. The yellow box marks the location ofrepresentative segments shown in B-G. (B) Wild-type expression ofmib2 RNAis confined to the FCs. (C) Expression of the Mib2-lacZ reporter recapitulatesthe RNA expression. (D) In Twi-Gal4>UAS-Rasact embryos, Mib2-lacZexpression expands throughout the visceral mesoderm. Bin-negative clustersin the foreground are somatic mesoderm. In contrast, mib2 and Mib2-lacZexpression remain restricted to the cells corresponding to the FCs, as in wildtype, in both a pnt null (E) and an activated pnt (F) background. Similarly, Mib2-lacZ expression retains a wild-type pattern in aop null (G) and aop activated(the constitutively repressing yanact; H) backgrounds.

2

RESEARCH ARTICLE Development (2019) 146, dev169003. doi:10.1242/dev.169003

DEVELO

PM

ENT

dumbfounded (duf )] and RhoGAP15B all appear normal in a pntact

background, whereas, as reported previously, expression of all threeexpands with pan-mesodermal expression of activated Ras (Rasact)(Fig. 3, Fig. S1; Leatherbarrow and Halfon, 2009; Lee et al., 2003).Double labeling with antibodies to Biniou, a general visceralmesoderm marker (Zaffran et al., 2001), showed that the observedexpansion is throughout the visceral mesoderm (Fig. 2D, Fig. 3D,E,insets). This suggests that non-FCs (i.e. FCMs) have beenre-specified as FCs, rather than that there has merely beenincreased FC proliferation. Consistent with this, expression of theFCM marker Lmd behaves in the reciprocal fashion: Lmdexpression decreases in the visceral mesoderm with Rasact

expression, but is unaffected in pnt null or pntact backgrounds(Fig. 3B,E,H; data not shown; Popichenko et al., 2013). Expressionof both Org-1 and RhoGAP15B is also unchanged in the pnt null,aop null, and yanact backgrounds (Fig. S1). Taken together, ourresults suggest that although Ras activity is sufficient to induce FCfates throughout the visceral mesoderm, neither pnt nor aop appearto play a significant role in this process.

Visceral founder cell expression of mib2 is repressedthrough ETS-type binding sites in the mib2 FC enhancerAlthough mib2 expression in the visceral mesoderm is notdependent on either pnt or aop, the mib2_FCenhancer enhancerwas identified in part based on the presence of ETS-type, putativePnt binding sites (Philippakis et al., 2006). We showed previouslythat mutation of a set of seven ETS sequences in this enhancercaused an expansion of reporter gene expression driven bythe mutated enhancer (Fig. 4G-I; Halfon et al., 2011). Like theexpression observed with activation of the Ras pathway, theexpanded reporter gene expression extends throughout the visceralmyoblast population as marked by Bin expression (data not shown).Interestingly, reporter gene expression is stronger in the FCs than in

the rest of the myoblasts (Fig. 4I, Fig. S2A-A″; data not shown).Activated Ras expression restores full-strength reporter activitysimilar to what is observed with the wild-type enhancer (Fig. 4J,Fig. S2B-B″; data not shown). However, as expected, the samedisparity in reporter expression between FCs and non-FCs as seen inthe wild-type background is seen with activated Pnt, which by itselfdoes not lead to expanded mib2 expression (Fig. 4K, Fig. S2C-C″;data not shown).

The expanded reporter gene expression observed upon mutationof the ETS sequences suggested that rather than being required forpositively activating mib2 expression – as expected based onanalogy to the requirement for Ras pathway signaling mediated byETS-family TFs in the somatic mesoderm (Halfon et al., 2000) –mib2 is repressed via TF binding at these sites. In order tounderstand better the nature of this repression, we decided tocharacterize the mib2 regulatory sequences more thoroughly.

Using sequence conservation with other Drosophila species as aguide, we first tested reporter gene activity with a truncated version ofthe mib2_FCenhancer containing a 5′ 120 bp deletion(Fig. 4A,B, Fig. S3). The deleted region includes one of theputative Pnt-binding sites previously mutated (‘site 1’; Fig. 4B,Fig. S3), as well as a non-canonical Pnt site suggested by protein-binding microarray experiments (Fig. S3; ‘siteN’; personalcommunication from Alan Michelson, NHLBI, Bethesda, MD,USA). The resulting mib2_FC626 enhancer has activity identical tothe longer mib2_FCenhancer (Fig. 4C-F), responds to Rasact andPntact ectopic expression in the same manner (Fig. S1E,F;Leatherbarrow and Halfon, 2009), and shows a similar Rasact-likeexpansion of reporter gene expression when the remaining six ETS-type sequences aremutated (constructmib2_FC626ETS; Fig. 4G-I). Incontrast, a more extensive 5′ 413 bp deletion (Fig. 4B, arrowhead;Fig. S3, gray arrow) leads to a complete loss of visceral mesodermactivity and only a limited residual expression elsewhere (data not

Fig. 3. Expression of multiple visceral FC and FCM markers respond to Ras but not to Pnt. (A-I) Stage 11 embryos that are either wild type (A-C), havepan-mesodermal expression of activated Ras (D-F; twi>Rasact), or have pan-mesodermal expression of activated Pnt (G-I; twi>pntact) were stained for FCand/or FCM markers. Consistent with results assessing mib2 expression, Ras activation led to increased FC and decreased FCM populations, whereas Pntactivation had no effect. A, D and G show expression of Org-1, an FC marker. Arrows indicate somatic mesoderm Org-1 expression, which expands in both theRasact and pntact backgrounds, in contrast to the visceral mesoderm expression, which expands only with Rasact. The embryo in A is slightly older than thosein B and C, leading to amore rounded, smoother pattern of visceral mesoderm expression. The inset in D is stained for the visceral mesodermmarker Bin in greenand Org-1 in magenta, demonstrating that the expansion of Org-1 expression is in the visceral mesoderm, as previously reported (Lee et al., 2003). B, E andH depict the FC marker kirrerp298-PZ, an enhancer-trap in the kirre (duf ) locus (green), and FCM marker Lmd (magenta). Ras activation causes conversion ofvisceral FCMs to FCs, resulting in the loss of Lmd with concomitant gain of Kirre expression; differences in the somatic mesoderm are more modest (arrows).The inset in E shows Kirre-lacZ expression in green and Bin expression in magenta, confirming the visceral mesoderm identity of the expanded Kirre-lacZ-expressing cells. C, F and I show in situ hybridization to RhoGAP15B RNA. All embryos are oriented dorsal up and anterior to the left.

3

RESEARCH ARTICLE Development (2019) 146, dev169003. doi:10.1242/dev.169003

DEVELO

PM

ENT

shown). As the mib2_FC626 enhancer behaves in all aspects like theoriginal mib2_FCenhancer, we used this shorter sequence as atemplate for further characterization of mib2 regulation.

We mutated the six remaining ETS-type sequences individuallyto determine which putative binding sites were responsible for theexpanded reporter gene expression (as sites 5 and 6 are closetogether, we treated them initially as a single site, 5-6). Expandedreporter gene expression is observed only with the site 5-6 pairedmutation (Fig. 4L, Fig. S4). We therefore further dissected this pairto test its component individual sites. Mutation of site 5 leads toexpansion ofmib2_FC626 enhancer activity throughout the visceralmesoderm (Fig. 4M), whereas site 6 by itself has a barely observablephenotype with expanded expression almost indistinguishable frombackground (Fig. S2G). The expanded expression observed with thesingle site 5 mutation is notably weaker than that observed withthe paired site 5-6 mutation, which itself has weaker expression thanthemib2_FC626ETS six-site-mutated enhancer (compare Fig. 4I,L,M).Although site 5 therefore appears to be the most crucial sitecontributing to expanded mib2_FC626 expression, the strongerexpression seen when multiple sites are mutated suggests that theseother sites are functional as well and contribute to overall enhanceractivity. Consistent with their more essential roles, site 5 is the mosthighly conserved of the six ETS sites in the mib2_FC626 sequence,followed by site 6 (Fig. S3B).

In addition to expanded visceral mesoderm expression, themib2_FC626ETS construct drives ectopic reporter gene expressionin the midline of the ventral nerve cord (Fig. 4M-O; Halfon et al.,2011). This ectopic midline expression is also observed with themib2_FC626site7 construct (Fig. 4N), but not with any of the other

Fig. 4. Mutagenesis of the mib2 FC enhancer reveals repression actingvia ETS binding sites. (A) Schematic of the mib2 locus. The location of theintronic mib2_FCenhancer regulatory sequence is indicated in gray. (B) Themib2_FCenhancer regulatory sequence, with conservation shown below ingreen. The gray portion of the sequence is deleted in the mib2_FC626constructs. A red arrowhead marks the location of the region deleted in theinactive 413 bp 5′ deletion. Red bars numbered 1-7 indicate the positionsof the tested ETS binding sites. Conservation track shows the 27-insectPhastCons conservation from the UCSC Genome Browser. (C,D) The shortermib2-FC626 enhancer (magenta) has activity indistinguishable from theoriginal mib2_FCenhancer enhancer (green); a higher magnification view isshown in D. (E,F) Single-channel images of the composite shown in D.(G-I) Mutation of the six ETS binding sites in the mib2-FC626 enhancer(mib2-FC626ETS; green in G,H) causes an expansion of reporter geneexpression throughout the visceral mesoderm. Expression in the FCs isstronger than the ectopic FCM-domain expression, as seen most clearly in thesingle-channel image in I, where arrows indicate the FCs and arrowheads pointto the FCMs. (J) In contrast, pan-mesodermal Ras activation causes similarectopic expression, but reporter gene levels are consistent throughout thevisceral mesoderm (arrows point to the wild-type FC domain and arrowheadsto the FCM domain). (K) Expression of activated Pnt, however, resembles theexpression seen in a wild-type background (arrows and arrowheads as in I).(L) Mutation of ETS sites 5 and 6 (mib2-FC626site5-6) causes reporter geneexpression to expand into the FCM domain, but the expanded expression isconsiderably weaker than that seenwith the full six-sitemutation (comparewithI). (M) Mutation of site 5 alone (mib2-FC626site5) also causes a weak reportergene expansion. The yellow dashed line indicates the border of the FCMdomain, as assessed by two-color imaging for the pan-visceral mesodermmarker Biniou (not shown). (N) Mutation of site 7 (mib2-FC626sit7) has no effectin the visceral mesoderm (not shown), but leads to ectopic reporter geneexpression in the midline of the ventral nerve cord (arrows). (O,P,Q) Similarectopic expression is observed when all ETS sites are mutated (O,P, arrows;compare with the same locations marked by arrows with the wild-typeenhancer in Q). (R) Similar ectopic reporter gene expression in the ventralmidline is also seen with the wild-type enhancer in an aopmutant background(arrows).

4

RESEARCH ARTICLE Development (2019) 146, dev169003. doi:10.1242/dev.169003

DEVELO

PM

ENT

single-site enhancer mutations. Similar ectopic mib2 expression isobserved in aop null mutant embryos (Fig. 4R), suggesting thatalthough Aop does not regulate mib2 in the visceral mesoderm, itmay act via this site to repress mib2 in the nervous system.

The de-repressed mib2 enhancer does not require Raspathway signaling for its activityIn the somatic mesoderm, we showed previously that Ras pathwayactivity is absolutely required for FC gene expression; for example,in the absence of both pnt and aop expression the eve_MHE FCenhancer is inactive, and mutation of the common ETS-type Pnt-and Aop-binding sites eliminates enhancer activity (Halfon et al.,2000). However, the de-repression of themib2 visceral FC enhancerseen with ETS site mutation suggested that Ras activity, normallynot present in the FCM population into which mib2 reporter geneexpression expands, might be dispensable when the mib2 enhanceris de-repressed. To test this, we analyzed the activity of the wild typeand mutated mib2_FC626 enhancers in a jeb null background,which lacks Ras signaling in the visceral mesoderm. Whereas thewild-type mib2_FC626 enhancer is inactive in the visceralmesoderm of jeb null embryos (Fig. 5A,C), the mib2_FC626ETS

mutated enhancer is expressed throughout the visceral mesodermjust as in a wild-type background (Fig. 5B,D, arrows). Staining forthe activated, di-phosphorylated form of MAPK confirmed that nosignaling was present in the jeb null background (Fig. 5F). Unlike inthe somatic mesoderm, therefore, Ras signaling is not requiredfor expression of a visceral FC gene in the absence of ETS site-mediated repression.

Visceral FC specification can occur in the absence of Raspathway signalingThe expanded expression of an FC gene throughout the visceralmesoderm we observed in the case of the de-repressedmib2_FC626 enhancer is reminiscent of what has been observedupon loss of function of the FCM transcriptional activator Lmd.Popichenko et al. (2013) showed that in lmd null embryos, FCmarkers such as org-1, hand and kirre expand throughout thevisceral mesoderm, with reciprocal loss of FCM genes such asVrp1. This resembles the phenotype observed with Ras activation.In a similar fashion, lmdmutation leads to the conversion of a smallsubset of FCMs to adult muscle precursor and pericardial cells(Sellin et al., 2009). However, these phenotypes are in sharpcontrast to what is observed for the bulk of the somatic mesoderm,where lmd loss of function has no effect on FC specification (Duanet al., 2001; Ruiz-Gomez et al., 2002). Interestingly, the FCM-specific gene sticks and stones (sns) remains expressed in the lmdvisceral mesoderm, suggesting that the observed FCM→FCconversion is incomplete (Estrada et al., 2006; Ruiz-Gomezet al., 2002). Given our results with the mib2 enhancer, wewondered whether lmd loss of function-induced FCM→FC re-specification could also occur in the absence of Ras pathwayactivity. Therefore, we tested jeb;lmd double-mutant embryos for arange of FC markers including org-1, mib2 and RhoGAP15B,which are expressed in all FCs, as well as Connectin (Con) andwingless (wg), which are expressed in only a subset of FCs (Bilderand Scott, 1998). In all cases, lmdwas fully epistatic to jeb: whereasin jeb null embryos no FCmarkers are expressed, expression in jeb;lmd embryos consistently resembles lmd alone, in most cases withexpanded FC expression (Fig. 6). Surprisingly, mib2 andRhoGAP15B, which expand throughout the visceral mesodermboth with Ras activation and upon mutation of the mib2 FCenhancer, do not appear to have expanded expression in the lmd

background when assayed by in situ hybridization (Fig. 6B,F). Thismay be indicative of an incomplete conversion of FCMs to FCs,consistent with the maintenance of sns expression in the FCMregion reported previously. Likewise, Wg expression also onlyexpands to a few additional cells and not throughout the entire widthof the anterior PS8 visceral mesoderm (Fig. 6S). On the other hand,we found that the mib2_FC626 reporter construct does show fullyexpanded expression in lmd embryos (Fig. 6B′). As expressiondriven by the mib2_FC626 enhancer appears identical toendogenous mib2 expression in all other contexts we haveexamined, it may be that the lack of expanded mib2 expressionobserved in lmd embryos simply represents a failure of detection,given that, similar to what we saw with the reporter construct inother backgrounds, reporter gene expression in the FCM region isweaker than that in the native FC region (Fig. S2D). Consistent withthis interpretation, we see a modest widening of mib2 expression inthe jeb;lmd embryos (Fig. 6D). Importantly, regardless of the exactdegree of expanded expression caused by lmd loss of function, FCexpression of all tested genes is clearly rescued in the jeb;lmdbackground (Fig. 6D,H,L,P), demonstrating the ability for Rassignaling to be bypassed in the absence of lmd expression.

To ensure that the rescue of FC specification observed in jeb;lmdmutants is not the result of a cryptic Ras signaling pathway activatedby loss of lmd, we checked for the presence of activated MAPK inthe double-mutant embryos. As expected when jeb is absent, no

Fig. 5. The mutated mib2 enhancer is active even in the absence of Rassignaling. (A-D) The wild-type (mib2-FC626WT; A,C) and ETS-site mutated(mib2-FC626ETS; B,D) mib2 enhancers were crossed into either a wild-type(A,B) or a jeb null background that lacks Ras signaling (C,D) in the visceralmesoderm and assessed for reporter gene expression. Black arrows indicatesomatic mesoderm expression, whereas white arrows point to visceralmesoderm expression. Arrowheads in B indicate the expanded reporter geneexpression, which, as shown previously in Fig. 4 and Fig. S2, is weaker in theexpanded region than in the normal region. Insets show a close-up of twosegments, with the domain of reporter gene expression outlined. In the jeb nullbackground (C), trunk visceral mesoderm expression is absent for the wild-type enhancer, with only somatic mesoderm (arrows) and caudal visceralmesoderm (arrowheads) activity still present. However, robust visceralmesoderm activity resembling that seen in a wild-type background is observedwith the mutated enhancer (B, arrows). This expression resembles that seenwith the wild-type enhancer following ectopic Ras activation (compare withFig. 2D). (E,F) Staining for activated MAPK (dpMAPK) shows crescents ofvisceral mesoderm expression in wild-type embryos (E, arrows), which areabsent in jeb null embryos (F). Embryos are oriented ventral up and anterior tothe left.

5

RESEARCH ARTICLE Development (2019) 146, dev169003. doi:10.1242/dev.169003

DEVELO

PM

ENT

activated MAPK is observed, regardless of presence or absence oflmd (Fig. 6X,Y).

DISCUSSIONBoth somatic and visceral muscle development require as an initialstep the specification of individual muscle founder cells from withinthe general myoblast pool. Superficially, the process appears alikefor both tissues: FCs fail to form in the absence of RTK/Ras/MAPKsignaling, and ectopic activation of the Ras pathway causes FCMsto be re-specified as FCs. A striking aspect of our current results isthat these seemingly similar events are brought about in amechanistically opposite fashion in the somatic versus visceralmesoderms. Our work therefore serves to underscore how commondevelopmental outcomes can derive from dramatically differentgene regulatory mechanisms.In the somatic mesoderm, it has been well-established that

positive induction via Ras/MAPK signaling is essential forspecifying FC fates (Buff et al., 1998; Carmena et al., 2002,1998; Halfon et al., 2000). In the visceral mesoderm, however,repression has primacy over induction. We demonstrate here thatRas/MAPK signaling acts in presumptive FCs to relieverepression of the FC fate, while Popichenko et al. (2013)previously established that it serves to prevent activation of FCMgenes. The primary activator of FCM genes is Lmd, which priorto FC specification is expressed in all visceral myoblasts

(Ruiz-Gomez et al., 2002; Popichenko et al., 2013). Ras/MAPK signaling in FCs causes phosphorylation of Lmdfollowed by its export from the nucleus and its degradation,preventing it from activating FCM-specific genes such as Vrp1(Popichenko et al., 2013). What happens at FC gene loci,however, had not previously been determined. Our results withthe mib2 enhancer demonstrate that FC genes can be activated inthe absence of Ras/MAPK signaling, through loss of repressorbinding at the enhancer.

A model for FC fate specificationThe simplest model, taking into account our results and those ofPopichenko et al. (2013), would be for Lmd to function as both the FCgene repressor and the FCM gene activator; loss of Lmd bindingfollowingRas/MAPK signalingwould thus simultaneously de-repressFC genes while halting activation of FCM genes. Although Lmd istypically viewed as an activator, some evidence suggests that it mayalso be capable of transcriptional repression (Cunha et al., 2010).However, chromatin immunoprecipitation studies have repeatedlyfailed to detect appreciable Lmd binding in the mib2 enhancer region(Busser et al., 2012; Cunha et al., 2010), and themib2 enhancer lacksgood candidate Lmd-binding sites – particularly in the crucial site 5-6region (M.S.H., data not shown) – even when surveyed using a rangeof binding motifs derived from multiple sources (Busser et al., 2012;Nitta et al., 2015; Zhu et al., 2011).

Fig. 6. FC gene expression is expanded in lmd and jeb;lmdmutant embryos. (A-Y) Expression of the FCmarkersmib2 (A-D),RhoGAP15B (E-H), Org-1 (I-L),Con (M-P), Wg (Q-U) and activated MAPK (V-Y) was assessed in wild type, lmd, jeb and jeb;lmd double-mutant embryos. Embryos in A-P and V-Y areoriented ventral up and anterior to the left and are shown as either whole embryos or as three-segment close-ups (two segments in V-Y). Q is oriented withventral to the bottom and anterior to the left; the arrow points to the region shown in close up in R-U. (A) In situ hybridization for mib2 RNA. (A′) Reporter geneexpression driven by themib2_FC626 enhancer. (B) In the lmdmutant background,mib2RNA expression resembles wild type, but the reporter construct (B′) hasexpanded expression similar to that seen with Ras activation or ETS-site mutation. (C,D) Visceral mesoderm expression of mib2 is absent in jeb embryos (C)but is restored and mildly expanded in the jeb;lmd background (D). (E-H) RhoGAP15B expression (E) does not show expansion in lmd embryos (F), but islikewise absent in jeb (G) and restored in a jeb;lmd (H) mutant background. (I-L) Org-1 expression expands in lmd (J), is lost in jeb (K), and is restored in jeb;lmd (L)embryos. (M-P) The same is true for Con, although expression remains confined to its wild-type anterior-posterior domain. (Q) Wg is expressed in a singlevisceral muscle FC in the wild-type stage 11 embryo (arrow). Higher magnification views reveal that the cell number is increased in lmd (S, arrows) and jeb;lmd(U, arrows) mutant embryos, but Wg expression is absent in the jeb null background (T). (V-Y) Staining for activated (diP) MAPK confirms that MAPKactivation is normal in lmd embryos (W) but absent in the jeb (X) and jeb;lmd (Y) visceral mesoderms. Arrows indicate visceral mesoderm expression,asterisks mark MAPK activation in the tracheal pits.

6

RESEARCH ARTICLE Development (2019) 146, dev169003. doi:10.1242/dev.169003

DEVELO

PM

ENT

We therefore favor a basic model in which Lmd serves as anactivator of the FC gene repressor, such that loss of Lmd leads toloss of repressor activity in FCs and subsequent expression of FCgenes (Fig. 7). In wild-type embryos, the main role of MAPKsignaling is thus to cause phosphorylation and degradation of Lmd,whereas in lmd mutant embryos, MAPK signaling becomesirrelevant as Lmd is already absent. The repressor could also be adirect target of MAPK, leading to its rapid displacement upon theonset of MAPK signaling (e.g. similar to what happens with Aop atother loci; Rebay and Rubin, 1995). This would be consistent withthe rapid time course of FC fate specification following MAPKactivation. We surmise that there are also additional, still unknownFC gene activators, activity of which may or may not be MAPKdependent. Tests of these various refinements to the basic modelwill require identification of the FC gene repressor.In the somatic mesoderm, the repressor Tramtrack69 (Ttk69; Ttk)

appears to play a role as an lmd-dependent FC gene repressor similarto what we posit here for visceral FC fate repression. ttk69expression is activated downstream of lmd in FCMs, where itrepresses the transcription of FC genes (Ciglar et al., 2014). In FCs,Ttk69 is likely post-translationally degraded in a Ras-dependentmanner (Ciglar et al., 2014; Li et al., 1997), relieving repression ofFC genes concurrent with Ras-dependent relief of Aop-mediatedrepression and induction via Pnt and/or other activators. However,different mechanisms appear to be at work in the visceralmesoderm. Although ttk69 mutants do display some alteredvisceral mesoderm gene expression (Ciglar et al., 2014), visceralFCs appear to be correctly specified and FC-specific genes such asmib2 are expressed in the appropriate pattern, without expansioninto the FCM field (Ciglar et al., 2014; S.E.P., unpublishedobservations).

A balance of ETS factors?Given the de-repression phenotypes observed on mutation of theETS sites in the mib2 enhancer, we favor the likelihood that therelevant repressor is a member of the ETS TF family, either by itselfor working redundantly with Pnt and/or Aop. Several other ETS-domain proteins exist in Drosophila (Chen et al., 1992), althoughtheir expression patterns and mutant phenotypes are for the mostpart not well defined. A role for additional ETS proteins was alsopreviously suggested for the dorsal somatic mesoderm, where pntloss of function leads to a partial loss of Eve-expressing FCs, butmutation of ETS binding sites completely eliminates expressiondriven by the eve_MHE enhancer (Halfon et al., 2000). Althoughour data argue against an absolute requirement for either Pnt or Aop,we cannot rule out a more limited contribution from these factors,and their requirement for expansion of FC fates (as opposed togeneration of normal FC fates) has not been fully tested. Indeed,

chromatin immunoprecipitation experiments indicate that Pnt canbind to the mib2 enhancer region, although it is not known in whatcell types (Webber et al., 2018), and aop mutants show an effect onmib2 expression in the ventral midline (Fig. 4R).

Although there is considerable evidence demonstrating therequirement for RTK/Ras/MAPK signaling for somatic FCspecification, the molecular details on the mechanisms governingMAPK-dependent activation and repression come mainly fromstudies of a single transcriptional enhancer, eve_MHE (BoisclairLachance et al., 2018; Halfon et al., 2000; Webber et al., 2018,2013). It is clear that ETS factor-dependent activation is essential forthe activity of this enhancer, as mutation of the major ETS bindingsites renders the enhancer non-functional (Halfon et al., 2000).However, recent studies suggest that instead of an abrupt switchbetween activation and repression due to mutually exclusiveenhancer occupancy by Pnt and Aop, there is a more subtlebalance between these TFs and their binding to the multiple high-and low-affinity ETS binding sites found in the enhancer (BoisclairLachance et al., 2018; Webber et al., 2018). Other somaticFC enhancers have not been rigorously tested with respect toETS-family binding, and it may be that the trade-off betweenactivation and repression differs among them. This would help toexplain the results of Buff et al. (1998), who demonstrated thatdifferent FCs are specified at different levels of RTK/Ras signaling.One way to achieve such differential sensitivity would be through amixture of activating versus repressing ETS TFs competing forbinding at a range of high- and low-affinity sites. Such a mechanismcould provide for exquisitely fine-tuned responsiveness to Ras/MAPK signaling, making this an appealing possibility.

In this vein, we note that our detailed molecular insights forvisceral FCs again come mainly from the study of a single enhancer,mib2_FC626. Here, elimination of the major ETS binding sitesleads to increased activity, the opposite of the situation with thesomatic eve_MHE enhancer. However, preliminary analysis of othervisceral FC enhancers suggests that eliminating ETS binding sitescan lead to loss of enhancer function, more similar to what is seen inthe somatic musculature (Y.Z., unpublished observations). Thus,although our data from the mib2_FC626 enhancer as well as fromanalysis of lmdmutants clearly establishes de-repression rather thaninduction as the major role for Ras pathway signaling duringvisceral FC specification, it may be that the molecular basis for howthis signaling is modulated by ETS-family TFs at the enhancer levelis complex and balanced individually at each FC gene enhancer.Taken together, these plus other recent results (Boisclair Lachanceet al., 2018; Webber et al., 2018) point to an elaborate interplaybetween Ras signaling, ETS TFs, and subtly tuned binding sites,and highlight the need for detailed molecular studies of a morecomprehensive set of both somatic and visceral FC enhancers.

Fig. 7. A model for visceral foundercell specification. In FCMs (left), Lmdactivates FCM-specific genes as wellas an FC-gene repressor, which keepsFC-specific genes shut off. In FCs(right), activation of MAPK leads to thedegradation of Lmd, preventingactivation of both the FCM genes andthe FC-gene repressor. MAPK mayalso act directly on the FC-generepressor (dashed line). Loss ofrepression allows for expression of theFC genes, possibly in conjunction withadditional activators (not shown).

7

RESEARCH ARTICLE Development (2019) 146, dev169003. doi:10.1242/dev.169003

DEVELO

PM

ENT

MATERIALS AND METHODSDrosophila strains and geneticsOregon-R was used as the wild type. Mutant stocks are described inFlyBase (Gramates et al., 2017) and include pntΔ88, aop1, lmd1 and jeb576

(Weiss et al., 2001). mib2_FCenhancer-lacZ has been described byPhilippakis et al. (2006) and the rp298-lacZ (FlyBase: kirrerp298-PZ) linewas used to assess kirre (duf ) expression (Nose et al., 1998). Ectopicexpression was achieved using the Gal4-UAS system (Brand andPerrimon, 1993) and used lines Twi-Gal4 (FlyBase: P{Gal4-twi.G})(Greig and Akam, 1993), UAS-Ras1Act (Carmena et al., 1998),UAS-PntP2VP16 (Halfon et al., 2000) and UAS-yanAct (Rebay andRubin, 1995). Mutant lines were rebalanced over lacZ-marked balancers toallow for genotyping of embryos.

Immunohistochemistry and microscopyFor all analyses, a minimum of ten embryos were analyzed in detail, withrepresentative images chosen for publication. All reported phenotypes werefully penetrant. Antibody staining was performed using standardDrosophila methods (Müller, 2008). The following primary antibodieswere used: mouse anti-β-galactosidase (Promega, Z3783; 1:500), rabbitanti-GFP (Abcam, ab290; 1:10,000), rabbit anti-Bin (gift of Eileen Furlong,EMBL, Heidelberg, Germany; 1:300), rabbit anti-Lmd (gift of HanhNguyen, Friedrich-Alexander-University of Erlangen-Nürnberg, Erlangen,Germany; 1:1000), rat anti-Org-1 (gift of Manfred Frasch, Friedrich-Alexander-University of Erlangen-Nürnberg; 1:250), mouse anti-activatedMAPK (diphosphorylated ERK1&2, Sigma, M9692; 1:250; fixed in 8%paraformaldehyde), mouse anti-Wg (4D4, Developmental StudiesHybridoma Bank; 1:100), mouse anti-Con (C1.427, DevelopmentalStudies Hybridoma Bank; 1:250). ABC kit (Vector Laboratories) wasused for immunohistochemical staining. Differential interference contrast(DIC) microscopy was performed using a Zeiss Axioskop 2 microscope andOpenlab (PerkinElmer) software for image capture. The followingsecondary antibodies were used for fluorescent staining: anti-mouseAlexa Fluor 488 (Molecular Probes, A11029; 1:250), anti-mouse AlexaFluor 633 (Molecular Probes, A21052; 1:500), anti-rabbit Alexa Fluor 488(Molecular Probes, A11034; 1:250), anti-rabbit Alexa Fluor 633 (MolecularProbes, A21071; 1:500), anti-rat Alexa Fluor 633 (Molecular Probes,A21094; 1:500). Fluorescent staining was visualized by confocalmicroscopy using a Leica SP2 confocal microscope. In situ hybridizationfor detection of mib2 and RhoGAP15B transcripts was as previouslydescribed (Leatherbarrow and Halfon, 2009). Color and brightness ofacquired images were adjusted using Adobe Photoshop.

Site-directed mutagenesis and transgenesisThe mib2_FC626 reporter was made by digesting the mib2_FCenhancersequence with BamHI and cloning the resulting 3′ portion into eitherpattBnucGFPh (for GFP reporter expression) or pattB-nucLacZ (for lacZreporter expression). Mutagenesis of the mib2 enhancer was performed byoverlap-extension PCR (Ho et al., 1989), and sequences were cloned intopattBnucGFPh. Mutated sequences are shown in Fig. S3 (primer sequencesavailable in Table S1). Transgenic flies were generated by Genetic Services(Cambridge, MA, USA) using phiC31-transgenesis and the attP2 landingsite. The availability of both GFP and lacZ reporter versions of the wild-typemib2_FC626 enhancer allowed for easy comparison of wild-type andmutant enhancers by crossing a lacZ reporter line to a GFP reporter line anddouble labeling for both lacZ and GFP.

AcknowledgementsWe thank Manfred Frasch, Eileen Furlong, Alan Michelson, Hanh Nguyen, theBloomington Drosophila Stock Center (NIH P40OD018537) and the DevelopmentalStudies Hybridoma Bank (created by the NICHD of the NIH and maintained atThe University of Iowa, Department of Biology, Iowa City, IA 52242) for fly stocksand antibodies. Elizabeth Brennan, Sam Hasenauer, Qiyun Zhu and JackLeatherbarrow helped with experiments. Steve Gisselbrecht and Michael Buckprovided critical comments on the manuscript.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsConceptualization: Y.Z., M.S.H.; Investigation: Y.Z., S.E.P., E.D.; Writing - originaldraft: Y.Z., M.S.H.; Writing - review & editing: Y.Z., S.E.P.; Supervision: M.S.H.;Project administration: M.S.H.; Funding acquisition: M.S.H.

FundingThis work was supported by the American Cancer Society [grant RSG-09-097-01-DDC to M.S.H.] and by the Biochemistry Department of the University at Buffalo.

Supplementary informationSupplementary information available online athttp://dev.biologists.org/lookup/doi/10.1242/dev.169003.supplemental

ReferencesBilder, D. and Scott, M. P. (1998). Hedgehog and wingless induce metameric

pattern in the Drosophila visceral mesoderm. Dev. Biol. 201, 43-56.Boisclair Lachance, J.-F., Webber, J. L., Hong, L., Dinner, A. R. and Rebay, I.

(2018). Cooperative recruitment of Yan via a high-affinity ETS supersite organizesrepression to confer specificity and robustness to cardiac cell fate specification.Genes Dev. 32, 389-401.

Brand, A. and Perrimon, N. (1993). Targeted gene expression as a means ofaltering cell fates and generating dominant phenotypes. Development 118,401-415.

Buff, E., Carmena, A., Gisselbrecht, S., Jimenez, F. andMichelson, A. M. (1998).Signalling by the Drosophila epidermal growth factor receptor is required for thespecification and diversification of embryonic muscle progenitors. Development125, 2075-2086.

Busser, B. W., Huang, D., Rogacki, K. R., Lane, E. A., Shokri, L., Ni, T., Gamble,C. E., Gisselbrecht, S. S., Zhu, J., Bulyk, M. L. et al. (2012). Integrative analysisof the zinc finger transcription factor Lame duck in the Drosophila myogenic generegulatory network. Proc. Natl. Acad. Sci. USA 109, 20768-20773.

Carmena, A., Gisselbrecht, S., Harrison, J., Jimenez, F. and Michelson, A. M.(1998). Combinatorial signaling codes for the progressive determination of cellfates in the Drosophila embryonic mesoderm. Genes Dev. 12, 3910-3922.

Carmena, A., Buff, E., Halfon, M. S., Gisselbrecht, S., Jimenez, F., Baylies, M. K.and Michelson, A. M. (2002). Reciprocal regulatory interactions between theNotch and Ras signaling pathways in the Drosophila embryonic mesoderm. Dev.Biol. 244, 226-242.

Chen, T., Bunting, M., Karim, F. D. and Thummel, C. S. (1992). Isolation andcharacterization of five Drosophila genes that encode an ets-related DNA bindingdomain. Dev. Biol. 151, 176-191.

Ciglar, L., Girardot, C., Wilczynski, B., Braun, M. and Furlong, E. E. M. (2014).Coordinated repression and activation of two transcriptional programs stabilizescell fate during myogenesis. Development 141, 2633-2643.

Cunha, P. M. F., Sandmann, T., Gustafson, E. H., Ciglar, L., Eichenlaub, M. P.and Furlong, E. E. M. (2010). Combinatorial binding leads to diverse regulatoryresponses: Lmd is a tissue-specific modulator of Mef2 activity. PLoS Genet. 6,e1001014.

Dobi, K. C., Schulman, V. K. and Baylies, M. K. (2015). Specification of thesomatic musculature in Drosophila. Wiley Interdiscip. Rev. Dev. Biol. 4, 357-375.

Duan, H., Skeath, J. B. and Nguyen, H. T. (2001). Drosophila Lame duck, a novelmember of the Gli superfamily, acts as a key regulator of myogenesis bycontrolling fusion-competent myoblast development. Development 128,4489-4500.

Englund, C., Loren, C. E., Grabbe, C., Varshney, G. K., Deleuil, F., Hallberg, B.and Palmer, R. H. (2003). Jeb signals through the Alk receptor tyrosine kinase todrive visceral muscle fusion. Nature 425, 512-516.

Estrada, B., Choe, S. E., Gisselbrecht, S. S., Michaud, S., Raj, L., Busser, B. W.,Halfon, M. S., Church, G. M. and Michelson, A. M. (2006). An integratedstrategy for analyzing the unique developmental programs of different myoblastsubtypes. PLoS Genet. 2, e16.

Gramates, L. S., Marygold, S. J., Santos, G. D., Urbano, J.-M., Antonazzo, G.,Matthews, B. B., Rey, A. J., Tabone, C. J., Crosby, M. A., Emmert, D. B. et al.(2017). FlyBase at 25: looking to the future. Nucleic Acids Res. 45, D663-D671.

Greig, S. and Akam, M. (1993). Homeotic genes autonomously specify one aspectof pattern in the Drosophila mesoderm. Nature 362, 630-632.

Halfon, M. S., Carmena, A., Gisselbrecht, S., Sackerson, C. M., Jimenez, F.,Baylies, M. K. and Michelson, A. M. (2000). Ras pathway specificity isdetermined by the integration of multiple signal-activated and tissue-restrictedtranscription factors. Cell 103, 63-74.

Halfon,M. S., Zhu, Q., Brennan, E. R. and Zhou, Y. (2011). Erroneous attribution ofrelevant transcription factor binding sites despite successful prediction of cis-regulatory modules. BMC Genomics 12, 578.

Ho, S. N., Hunt, H. D., Horton, R. M., Pullen, J. K. and Pease, L. R. (1989). Site-directed mutagenesis by overlap extension using the polymerase chain reaction.Gene 77, 51-59.

Imperial, R., Toor, O. M., Hussain, A., Subramanian, J. and Masood, A. (2017).Comprehensive pancancer genomic analysis reveals (RTK)-RAS-RAF-MEK as a

8

RESEARCH ARTICLE Development (2019) 146, dev169003. doi:10.1242/dev.169003

DEVELO

PM

ENT

key dysregulated pathway in cancer: its clinical implications. Semin. Cancer Biol.S1044-579X(17)30275-4.

Leatherbarrow, J. R. and Halfon, M. S. (2009). Identification of receptor-tyrosine-kinase-signaling target genes reveals receptor-specific activities and pathwaybranchpoints during Drosophila development. Genetics 181, 1335-1345.

Lee, H.-H., Norris, A., Weiss, J. B. and Frasch, M. (2003). Jelly belly proteinactivates the receptor tyrosine kinase Alk to specify visceral muscle pioneers.Nature 425, 507-512.

Li, S., Li, Y., Carthew, R. W. and Lai, Z.-C. (1997). Photoreceptor cell differentiationrequires regulated proteolysis of the transcriptional repressor Tramtrack. Cell 90,469-478.

Martin, B. S., Ruiz-Gomez, M., Landgraf, M. and Bate, M. (2001). A distinct set offounders and fusion-competent myoblasts make visceral muscles in theDrosophila embryo. Development 128, 3331-3338.

Muller, H. A. (2008). Immunolabeling of embryos.Methods Mol. Biol. 420, 207-218.Nitta, K. R., Jolma, A., Yin, Y., Morgunova, E., Kivioja, T., Akhtar, J., Hens, K.,Toivonen, J., Deplancke, B., Furlong, E. E. M. et al. (2015). Conservation oftranscription factor binding specificities across 600 million years of bilateriaevolution. eLife 4, e04837.

Nose, A., Isshiki, T. and Takeichi, M. (1998). Regional specification of muscleprogenitors in Drosophila: the role of the msh homeobox gene.Development 125,215-223.

Philippakis, A. A., Busser, B. W., Gisselbrecht, S. S., He, F. S., Estrada, B.,Michelson, A. M. and Bulyk, M. L. (2006). Expression-guided in silico evaluationof candidate Cis regulatory codes for drosophila muscle founder cells. PLoSComput. Biol. 2, e53.

Popichenko, D., Hugosson, F., Sjogren, C., Dogru, M., Yamazaki, Y.,Wolfstetter, G., Schonherr, C., Fallah, M., Hallberg, B., Nguyen, H. et al.(2013). Jeb/Alk signalling regulates the Lame duck GLI family transcription factorin the Drosophila visceral mesoderm. Development 140, 3156-3166.

Rebay, I. and Rubin, G. M. (1995). Yan functions as a general inhibitor ofdifferentiation and is negatively regulated by activation of the Ras1/MAPKpathway. Cell 81, 857-866.

Ruiz-Gomez, M., Coutts, N., Suster, M. L., Landgraf, M. and Bate, M. (2002).myoblasts incompetent encodes a zinc finger transcription factor required tospecify fusion-competent myoblasts in Drosophila. Development 129, 133-141.

Schlessinger, J. (2000).Cell signaling by receptor tyrosine kinases.Cell103, 211-225.Sellin, J., Drechsler, M., Nguyen, H. T. and Paululat, A. (2009). Antagonistic

function of Lmd and Zfh1 fine tunes cell fate decisions in the Twi and Tin positivemesoderm of Drosophila melanogaster. Dev. Biol. 326, 444-455.

Tidyman, W. E. and Rauen, K. A. (2009). The RASopathies: developmentalsyndromes of Ras/MAPK pathway dysregulation. Curr. Opin. Genet. Dev. 19,230-236.

Webber, J. L., Zhang, J., Mitchell-Dick, A. and Rebay, I. (2013). 3D chromatininteractions organize Yan chromatin occupancy and repression at the even-skipped locus. Genes Dev. 27, 2293-2298.

Webber, J. L., Zhang, J., Massey, A., Sanchez-Luege, N. and Rebay, I. (2018).Collaborative repressive action of the antagonistic ETS transcription factorsPointed and Yan fine-tunes gene expression to confer robustness in Drosophila.Development 145, dev165985.

Weiss, J. B., Suyama, K. L., Lee, H.-H. and Scott, M. P. (2001). Jelly belly: aDrosophila LDL receptor repeat-containing signal required for mesodermmigration and differentiation. Cell 107, 387-398.

Zaffran, S., Kuchler, A., Lee, H. H. and Frasch, M. (2001). biniou (FoxF), a centralcomponent in a regulatory network controlling visceral mesoderm developmentand midgut morphogenesis in Drosophila. Genes Dev. 15, 2900-2915.

Zhu, L. J., Christensen, R. G., Kazemian, M., Hull, C. J., Enuameh, M. S.,Basciotta, M. D., Brasefield, J. A., Zhu, C., Asriyan, Y., Lapointe, D. S. et al.(2011). FlyFactorSurvey: a database of Drosophila transcription factor bindingspecificities determined using the bacterial one-hybrid system.Nucleic Acids Res.39, D111-D117.

9

RESEARCH ARTICLE Development (2019) 146, dev169003. doi:10.1242/dev.169003

DEVELO

PM

ENT