Current Biology, Vol. 15, 624–636, April 12, 2005, ©2005 Elsevier Ltd All rights reserved. DOI 10.1016/j.cub.2005.02.057

Distinct Functions of the Tribolium zerknulltGenes in Serosa Specification and Dorsal Closure

Maurijn van der Zee,1 Nicola Berns,2

and Siegfried Roth1,*1Institut fur EntwicklungsbiologieUniversität zu KölnGyrhofstraße 17D-50931 KölnGermany2 Institut fur ZellbiologieUniversität TubingenAbteilung Genetik der TiereAuf der Morgenstelle 28D-72076 TubingenGermany

Summary

Background: In the long-germ insect Drosophila, a sin-gle extraembryonic membrane, the amnioserosa, cov-ers the embryo at the dorsal side. In ancestral short-germ insects, an inner membrane, the amnion, coversthe embryo ventrally, and an outer membrane, the se-rosa, completely surrounds the embryo. An early differ-entiation step partitions the uniform blastoderm intothe anterior-dorsal serosa and the posterior-ventralgerm rudiment giving rise to amnion and embryoproper. In Drosophila, amnioserosa formation dependson the dorsoventral patterning gene zerknullt (zen), aderived Hox3 gene.Results: The short-germ beetle Tribolium castaneumpossesses two zen homologs, Tc-zen1 and Tc-zen2. Tc-zen1 acts early and specifies the serosa. The loss ofthe serosa after Tc-zen1 RNAi is compensated by anexpansion of the entire germ rudiment toward the ante-rior. Instead of the serosa, the amnion covers theembryo at the dorsal side, and later size regulation nor-malizes the early fate shifts, revealing a high degree ofplasticity of short-germ development. Tc-zen2 actslater and initiates the amnion and serosa fusion re-quired for dorsal closure. After Tc-zen2 RNAi, theamnion and serosa stay apart, and the embryo closesventrally, assuming a completely everted (inside-out)topology.Conclusions: In Tribolium, the duplication of the zengenes was accompanied by subfunctionalization. Oneof the paralogues, Tc-zen1, acts as an early anterior-posterior patterning gene by specifying the serosa. Inabsence of the serosa, Tribolium embryogenesis ac-quires features of long-germ development with a singleextraembryonic membrane. We discuss implicationsfor the evolution of insect development including theorigin of the zen-derived anterior determinant bicoid.

Introduction

One of the evolutionary innovations that probably con-tributed to the unparalleled success of the insects is

*Correspondence: siegfried.roth@uni-koeln.de

the formation of two protecting extraembryonic mem-branes: the amnion and the serosa [1]. These mem-branes are absent from all other arthropods [2, 3]. Theprominent distinction between cells of the presumptiveserosa and cells of the germ rudiment, giving rise tothe amnion and the embryo proper, is the first differenti-ation step that occurs within the uniform blastoderm.This stage is called the “differentiated blastoderm” [4].Remarkably, this stage is absent in a small group ofinsects, the higher dipterans, to which Drosophila be-longs [5, 6]. In Drosophila, the germ rudiment occupiesthe entire blastoderm. A single extraembryonic mem-brane, the amnioserosa, develops from a dorsal stripeof the blastoderm and covers the yolk at the dorsal sideof the egg during further development.

The red flour beetle Tribolium castaneum may be arepresentative of all other insects, given that it still pos-sesses an amnion and a serosa and the differentiatedblastoderm stage. At this stage, the prospective serosais visible as a dorsally tilted anterior cap of flatteningcells. During gastrulation, the serosa, together with theamnion, begins to cover the embryo proper at the pos-terior pole, forming the posterior amniotic fold. Thesame occurs to a lesser extent at the anterior pole,forming the anterior amniotic fold. When the crests ofthe amniotic folds meet, a round serosal window isformed. Finally, the serosal window closes, forming acontinuous outer membrane, the serosa, and an innermembrane, the amnion [7] (see also Figures 8A–8D forschematic representation).

In Drosophila, amnioserosa formation requires thegene zerknullt (zen), which encodes a homeobox tran-scription factor [8] and is a target gene of maternal andzygotic dorsoventral (DV) morphogen gradients. Duringearly blastoderm stages, the maternal NF-κB/Dorsalprotein gradient represses zen in the ventral half of theembryo and thus confines its transcription to a broaddorsal domain [9]. During gastrulation, a zygotic bonemorphogenetic protein (BMP) gradient with peak levelsalong the dorsal midline activates zen in a 5-cell-widedorsal expression domain comprising the cells thatgive rise to the amnioserosa (see [10] for review). Inzen mutants, the amnioserosa is lost and is replacedby dorsal ectoderm. zen maps to the Antennapediacomplex (ANT-C) between deformed and probos-cipedia [11]. This region harbors two closely linkedtranscription units, zen and z2, with identical expres-sion patterns [12]. Because zen alone could rescue adeletion covering both zen and z2, it was suggestedthat z2 is dispensible [12]. This was later confirmed byproducing a deletion specific for z2 [13].

zen homologs have also been found in Tribolium cas-taneum and the grasshopper Schistocerca gregaria.Schistocerca-zen is expressed in the serosa and lateamnion, whereas Tribolium zen is expressed in the se-rosa [14, 15]. Later sequencing of the Tribolium ANT-Crevealed two zen homologs: Tc-zen1, which corres-ponds to the homolog cloned by Falciani et al. [14], andTc-zen2 [16]. Notably, the duplication of zen in Dro-sophila and Tribolium has happened independently, as

Functional Analysis of the Tribolium zen Genes625

inferred from phylogenetic analysis [16] and the ab-sence of z2 in Drosophila pseudoobscura [17]. So far,expression of Tc-zen2 has not been investigated. Theexpression of Schistocerca zen and Tribolium zen1suggests a conserved function for zen in the patterningor differentiation of extraembryonic tissues in all in-sects. However, until now, no functional study has beenconducted except in Drosophila.

In this paper, we show that after duplication, the zengenes of Tribolium acquired partially different expres-sion patterns and diverged completely with regard totheir function. RNAi with Tc-zen2 results in incorrectdorsal closure, generating completely everted (inside-out) larvae. RNAi with Tc-zen1 leads to the completeloss of the anteriormost cell fate, the serosa, and acompensating expansion of the more posterior germrudiment. The fact that Tc-zen1 is involved in earlyspecification of an anterior structure might have been afavorable condition for the evolution of the zen-derivedanterior determinant bicoid, which specifies head andthorax in higher dipterans [18]. Despite absence of theserosa after Tc-zen1 RNAi, size regulation generatesnormal larvae that exhibit perfect dorsal closure. Thisdevelopmental plasticity sheds light on the evolution oflong-germ insects with a single extraembryonic mem-brane (the amnioserosa).

Results

Expression of the zerknullt Genes in TriboliumWe first asked whether the duplication of the zen genesin the lineage leading to Tribolium led to differences inthe expression pattern of the two paralogues. Tc-zen2is expressed similarly to Tc-zen1 [14] in the presump-tive serosa at the anterior pole of blastoderm embryos(Figure 1A). The flat cells of the presumptive serosa areeasily recognized by the wider-spaced nuclei (Figure1B). During and after gastrulation, when the serosa cov-ers the embryo, Tc-zen2 continues to be expressed inthe serosa, like Tc-zen1 (Figure 1C). As for Tc-zen1, no

Figure 1. Expression Pattern of Tc-zen2

All embryos show in situ hybridizations forTc-zen2.(A) Optical midsection of a differentiatedblastoderm embryo. Tc-zen2 is expressed ina tilted anterior cap.(B) Fluorescence image of the surface of theembryo shown in (A). DAPI staining visual-izes the nuclei. Tc-zen2 is expressed in thewider-spaced nuclei of the presumptiveserosa.(C) Serosal window stage. Tc-zen2 is ex-pressed in the serosa, which completelycovers the embryo. The inset shows the se-rosal surface at larger magnification. Tc-zen2transcripts are found in the cytoplasm sur-rounding the nuclei of the stretched-out se-rosal cells.(D) Extended germband stage. Tc-zen2 is ex-pressed in the amnion, covering the thoracicand posterior head region. The serosa hasbeen removed. Arrowheads indicate the lim-its of the expression. “sw” denotes serosalwindow, and “hl” denotes head lobe.

expression was detected in the embryo proper or inthe early amnion. Whereas Tc-zen1 expression remainsrestricted to the serosa, Tc-zen2 shows a new expres-sion domain in the late amnion (Figure 1D). Tc-zen2transcripts can be detected in the flattened amnioticcells that cover the posterior head and the anterior tho-rax region (Figure 1D). This expression is initiatedshortly before the amnion and serosa begin to fuse(see below).

Functional Analysis of the zerknulltGenes by Parental RNAiTo investigate the function of the Tribolium zen genes,we performed knockdown experiments with the paren-tal RNAi technique. As a negative control, female pupaewere injected with lacZ dsRNA. Aside from 13% unfer-tilized or undeveloped eggs, all developing young em-bryos from these mothers were indistinguishable fromthe wild-type (Figure 2, penultimate bar). After injectionof 0.1 �g Tc-zen1 or Tc-zen2 dsRNA, similar fre-quencies of unfertilized or undeveloped embryos wereobserved (Figure 2). The developing embryos, however,displayed distinct, specific phenotypes, which are de-scribed in the next sections. Control in situ hybridiza-tions confirmed the absence of Tc-zen1 or Tc-zen2transcripts after the respective dsRNA injection and theabsence of both transcripts after the combined injec-tions. Furtermore, we monitored the interdependenceof the paralogues. Tc-zen2 does not control Tc-zen1expression because normal Tc-zen1 expression wasdetected after Tc-zen2 RNAi. In contrast, the early ex-pression of Tc-zen2 is dependent on Tc-zen1 becauseno early Tc-zen2 transcripts were found after Tc-zen1RNAi. However, late amniotic expression of Tc-zen2could be detected after Tc-zen1 RNAi (data not shown),demonstrating that this late expression is independentof Tc-zen1 and that a nonspecific knockdown of Tc-zen2 transcripts by Tc-zen1 dsRNA injections can beexcluded.

Current Biology626

Figure 2. Phenotypic Frequencies for Paren-tal RNAi with the Two Tc-zen Homologs

Frequencies of phenotypes after Tc-zen1RNAi, Tc-zen2 RNAi, combined RNAi, and acontrol injection with lacZ dsRNA. Every firstbar represents the germband stages ana-lyzed (12–18 hr). Every second bar repre-sents the cuticles analyzed (>96 hr). StrongRNAi phenotypes are shown in red, moder-ate RNAi phenotypes in orange, wild-typephenotypes in green, unfertilized or undevel-oped eggs in black, and nonspecific effectsin blue. Tc-zen1 RNAi has an early effect; theabnormal phenotype is later restored (com-pare first and second bar). Tc-zen2 RNAi hasa late effect (compare third and fourth bar).The combined RNAi is similar to Tc-zen1RNAi. The following is a short description ofthe categories. Germband extension pheno-types after Tc-zen1 RNAi: Strong pheno-type = big head and no serosa; moderatephenotype = normal-appearing head and noserosa. Cuticle phenotypes after Tc-zen1RNAi: Moderate phenotype = unhatched,strongly curved larvae. Cuticle phenotypesafter Tc-zen2 RNAi: Strong phenotype =completely everted; moderate phenotype =partially everted. Combined injection: seeTc-zen1 RNAi. Note that after Tc-zen1 RNAi,all differentiated blastoderm embryos (6–9hr) lack the serosa and have an expandedgerm rudiment (see text). The first bar, how-ever, shows a slightly later stage (12–18 hr),in which 71% display dramatic conse-quences of this phenotype (red), but 14% arealready partially recovered (orange).

Loss of Tc-zen1 Causes a Transformation zmof Presumptive Serosa into Germ Rudiment

In wild-type embryos at the differentiated blastoderm ptstage, a clear distinction arises between the wider-

spaced serosal cells and the more densely spaced cells tsof the germ rudiment (Figure 1B). This difference is

even more pronounced at gastrulation onset, markedaby the formation of the primitive pit at the posterior pole

(Figure 3A). At this stage, the serosal nuclei become pelarger than those of the germ rudiment because they

undergo polyploidization. After Tc-zen1 RNAi, the em- habryos never partition the blastoderm into regions with

distinct nuclear spacing (Figure 2; Figure 3G). The uonuclear density throughout the embryo resembles that

of the germ rudiment. Thus, high nuclear densities are atalso found in the anterior third of the embryo, where

the serosa would form normally. At the anterior tip, a ahslight decrease in nuclear density is observed (Figure

3G; Figure 4J). However, even here the nuclei are more sadensely packed than in the presumptive serosa of a

wild-type embryo at the corresponding stage. Further-4more, in this region, the nuclear size does not increase

as in serosal nuclei, and there is no sharp boundary [oseparating this region from the remaining cells.

The high nuclear density throughout the embryo can dRbe explained by assuming that Tc-zen1 RNAi causes

a loss of the serosal cell fate. Almost all cells of the Tsblastoderm appear to adopt the fate of germ rudiment.

The slight decrease in nuclear density at the anterior uppole could be due to an incomplete knockdown of Tc-

en1 after dsRNA injection. However, the later develop-ent of Tc-zen1 RNAi embryos suggests another ex-lanation and is described below. To test the assump-ion that the germ rudiment expands at the expense ofhe serosa after Tc-zen1 RNAi, we analyzed the expres-ion of five marker genes.Tc-twist served as a ventral marker. In the wild-type

t the uniform blastoderm stage, Tc-twist is weakly ex-ressed in a ventral stripe along the entire anteropost-rior (AP) axis. Shortly before primitive-pit formation,owever, Tc-twist is excluded from the emerging serosand is strongly upregulated in the germ rudmiment (Fig-res 4A and 4B; [19, 20]). In Tc-zen1 RNAi embryos, webserved this strong upregulation along the entire APxis (Figures 4I and 4J). The expanded stripe of strongwi expression includes the slightly wider-spaced cellst the anterior tip, demonstrating that they do not be-ave as serosal cells do. The change in twi expressionuggests that the whole ventral side of the embryodopted the cell fate of the germ rudiment.The expression of the novel gene Tc006A12 (Figures

C and 4D; GenBank accession number CB33537421]) and of the homeobox gene Tc-orthodenticle1 (Tc-td1 [22]) was analyzed to investigate how the AP coor-inates of the germ rudiment change after Tc-zen1NAi. In wild-type embryos at the primitive-pit stage,c006A12 and Tc-otd1 are expressed in a wedge-haped domain, immediately posterior the serosa (Fig-res 4C and 4D; data not shown). The onset of ex-ression of an additional posterior Tc006A12 domain

Functional Analysis of the Tribolium zen Genes627

Figure 3. The Development of Embryos afterParental Tc-zen1 RNAi

(A–F) Wild-type embryos; lateral views areshown unless otherwise indicated. (G–L) Em-bryos of the corresponding age after Tc-zen1RNAi. (A–E and G–K) DAPI stainings. (D–Fand J–L) Engrailed antibody stainings.(A) Onset of gastrulation (primitive-pit stage).The nuclei of the serosa are larger and wider-spaced than those of the germ rudiment.(B) Serosal window stage. The amnion andthe serosa grow over the head and abdomenof the embryo proper.(C) Extending germband stage. After closureof the serosal window, the amnion and theserosa cover the embryo as continuousmembranes. The amnion (A) is easily visible.The serosa has been removed.(D) Extended germband stage. Amnion andserosa have been removed.(E) Ventral view of the third Engrailed stripeof an extending germband embryo.(F) Retracting germband stage.(G) After Tc-zen1 RNAi, the distinction be-tween the widely spaced nuclei of the serosaand the densely spaced nuclei of the germrudiment is absent. All nuclei are denselyspaced.(H) During gastrulation, cells condenseslowly to the ventral side. No serosal windowis visible.(I) More cells than in the wild-type contributeto the head, which develops slower. A struc-ture similar to the posterior amniotic foldforms.(J) The cells of the big head express En-grailed. The En stripes are laterally ex-panded and further apart in the anteropos-terior dimension. Development of the head isdelayed with regard to development of theabdomen.(K) Ventral view of the third Engrailed stripeof an extending germband embryo. Thestripe consists of 1.5× more cells, which areless densely spaced.(L) At the retracted germband stage, Tc-zen1RNAi embryos look rather normal. Arrow-heads point to the double extraembryonicmembrane, covering the abdomen of theembryo. The following abbreviations wereused: a, amnion; paf, posterior amniotic fold;pp, primitive pit; and sw, serosal window.

(Figures 4C and 4D) allows accurate staging of Tc-zen1RNAi embryos in comparison to the wild-type. In Tc-zen1 RNAi embryos, the anterior Tc006A12 domain ex-pands and is shifted toward the anterior pole to coverthe area, which harbors the serosa in the wild-type (Fig-ures 4K and 4L). Because the expansion is more pro-nounced at the dorsal in comparison to the ventral side,the domain becomes more symmetrical along the DVaxis. The same expansion is observed for zygotic Tc-

otd1 expression (data not shown). These data suggestthat the serosa is replaced by enlarged anterior regionsof the germ rudiment.

The pair-rule genes Tc-eve [23] and Tc-hairy [24] wereused as markers to investigate whether more-posteriorfates of the germ rudiment expand as well. In the wild-type, Tc-eve is expressed in a broad posterior cap andin one anterior stripe at the very onset of primitive-pitformation (Figures 4E and 4F). At this stage, both do-

Current Biology628

mRtgotwditagtd

egRwt(iesswpaD

otdkempmmlt

AaIcrp

(Figure 4. Expansion of the Germ Rudiment after Tc-zen1 RNAi

((A–F) Wild-type differentiated blastoderm stages (A–F) and earlyagastrulation (G and H). Lateral views are shown unless indicatedaotherwise. (I–P) Corresponding stages after Tc-zen1 RNAi. (B), (D),s(E), and (G) and (J), (L), (N), and (P) show DAPI counterstainings of(the embyos shown above.d(A and B) Ventral view. Tc-twist is expressed in the germ rudiment,dbut not in the flattened serosa cells, characterized by wider-(spaced nuclei.p(C and D) Tc006A12 expression in the anterior germ rudiment, justaposterior of the serosa. Onset of expression in the primitive pit(allows accurate staging of the embryos.a(E and F) Tc-evenskipped at the onset of the differentiated blasto-pderm stage.

osterior amniotic fold (Figure 3B; see also Figure 8 for

G and H) Tc-hairy expression at the onset of gastrulation.I and J) Ventral view. Tc-twist upregulation has expanded to thenterior tip. Some slightly wider-spaced nuclei are present at thenterior tip. However, they express twi and are therefore not oferosal origin.K and L) The anterior domain of Tc006A12 expression shows aramatic expansion toward the anterior. Also, the more posterioromain seems to be expanded.

M and N) Tc-eve expression at the primitive-pit stage. Both theosterior cap and the stripe are expanded toward the anterior andre enlarged. No serosal nuclei are present.

O and P) All three Tc-hairy stripes expanded toward dorsal andnterior and shifted anteriorly. The effect is diminished for the moreosterior stripes.

ains are expanded and shifted anteriorly after Tc-zen1NAi (Figures 4M and 4N). The same was observed for

he expression of Tc-hairy (data not shown). Thus, allerm-rudiment cell fates, including the more-posteriornes, appear to be expanded toward the anterior. Fur-hermore, the Tc-eve stainings allowed us to assesshen Tc-zen1 starts to act. At the early uniform blasto-erm stage, Tc-eve is expressed in a posterior cap, with

ts anterior border in the middle of the blastoderm. Athis stage, no dramatic differences between wild-typend Tc-zen1 RNAi embryos could be detected, sug-esting that the fate shifts toward the anterior do notake place until shortly before the differentiated blasto-erm stage (data not shown).At the onset of gastrulation, Tc-eve and Tc-hairy are

xpressed in three stripes, restricted to the condensingerm rudiment (Figures 4G and 4H for Tc-hairy). Tc-zen1NAi causes an expansion and a shift of all stripes to-ard the anterior. This effect is most pronounced for

he anterior stripe and decreases toward the posteriorFigures 4O and 4P for Tc-hairy). Furthermore, Tc-hairys expressed along the entire DV axis in Tc-zen1 RNAimbryos. The same was observed for Tc-eve. At thistage, the expansion of the serosa has led to a con-iderable ventral condensation of the germ rudiment inild-type embryos. The absence of this process isrobably the main reason that the Tc-hairy stripes showdramatic dorsal expansion after Tc-zen1 RNAi (see

iscussion).In summary, these data demonstrate that the earliest

vert cell-differentiation step of the Tribolium embryo,he differentiation into serosal and germ rudiment cells,epends on Tc-zen1. The absence of serosal cells afternockdown of Tc-zen1 is compensated by an anteriorxpansion of the entire germ rudiment, including theore posterior fates. In absence of the serosa, both theosterior and ventral condensation of the germ rudi-ent are blocked. Concomitantly, the expression do-ains of early embryonic patterning genes cover a

arger area of the embryo surface. This leads to a situa-ion resembling long-germ development.

fter Loss of Tc-zen1, the Amnion Covers the Yolkt the Dorsal Side of the Embryon gastrulating wild-type embryos, the germ rudimentondenses to the ventral side, and the amnion and se-osa start to cover the abdomen by formation of the

Functional Analysis of the Tribolium zen Genes629

schematic representation). After formation of the fifthEngrailed (En) stripe, the amnion and the serosa havecompletely covered the embryo proper (Figure 3C showsonly the amnion). During later stages of Tc-zen1 RNAiembryos, cells also condense toward the ventral side,and an increasing number of wider-spaced nuclei be-comes visible at the anterior and dorsal side (Figures3H and 3I). However, this process is delayed in compar-ison to the wild-type, and no sharp boundary forms be-tween the area of the wider-spaced nuclei and the de-veloping germband. In addition, the abdomen starts tobe covered by a double layer of extraembryonic tissue(Figure 3I) that resembles the posterior amniotic foldof wild-type embryos. However, this process is muchdelayed, and a serosal window never forms. Theembryo will never be completely covered by extraem-bryonic cell layers, even when 17 En stripes are present(Figure 3J).

The appearance of widely spaced nuclei and adouble layer of extraembryonic tissue seems to indi-cate the presence of serosal cells. Three lines of evi-dence, however, strongly suggest that all extraembry-onic tissue in these embryos is of amniotic origin. First,cell divisions could still be detected by anti-phospho-histon3 (αPH3) staining among the wider-spaced nuclei(data not shown). Serosal cells never show αPH3 stain-ing because they stop dividing prior to gastrulation andlater become polyploid. In contrast, amniotic cellsmaintain mitotic activity even after they become flat-tened and widely spaced during wild-type development[20]. Second, the serosal marker Tc004A04 ([21]; GEN-BANK accession number CB335138) is not expressedin these flattened cells (n = 54; data not shown). Third,and most important, these cells express the amnioticmarker gene TcGATAx (Figure 5; [25]). In the wild-typeat the primitive-pit stage, TcGATAx expression is ex-cluded form the serosa. TcGATAx starts to be ex-pressed in a dorsal stripe of the germ rudiment and inthe primitive pit (Figure 5A). Weak expression is alsofound at the anterior rim of the germ rudiment just pos-terior to the serosa (marked by arrows in Figure 5A).Thus, early TcGATAx expression occurs presumably inthe amniotic anlagen. In accordance with this assump-

Figure 5. The Amnion Covers the Embryo atthe Dorsal Side of Tc-zen1 RNAi Embryos

(A and D) Wild-type embryos. (B and E) Em-bryos after Tc-zen1 RNAi. (C and F) DAPIcounterstainings of embryos shown in (B)and (E).(A) Primitive-pit stage. TcGATAx is expresseddorsally in the presumptive amnion and isabsent in the flattened cells of the serosa.On the surface, a faint rim of expression canbe observed, just posterior to the serosa (ar-rows; out of focus in lateral regions).(B and C) Tc-zen1 RNAi embryo at a laterstage than (A). At this stage, cells start toflatten at the anterior pole, as seen from thewider-spaced nuclei (C). These cells all ex-press TcGATAx.

(D) Extending germband stage. All cells of the amnion express TcGATAx. The arrowhead points to a small group of cells expressing TcGATAxwithin the head region.(E and F) After Tc-zen1 RNAi, all flattened cells express TcGATAx. The arrowhead points to the small group of cells expressing TcGATAxwithin the head region. The white lines in (F) demarcate the anterior and posterior end of the embryo proper.

tion, during germband extension, TcGATAx becomeshighly expressed in the amniotic cells that have foldedover the embryo at the ventral side (Figure 5D). In Tc-zen1 RNAi embryos, all wider-spaced, stretched-outcells that emerge anterior and dorsal to the condensingembryonic anlagen express TcGATAx (Figures 5B and5C). The expression level increases during further de-velopment, and the border toward the condensing em-bryonic anlagen sharpens (Figures 5E and 5F). TcGATAxis also expressed in the cells that form a double layerpartially covering the abdomen after Tc-zen1 RNAi (Fig-ure 5E; data not shown). Because no stretched-out cells,which lack TcGATAx expression, exist after Tc-zen1 RNAi,we conclude that all serosal cells are lost, and allstretched-out cells are amniotic. The flattening of thesecells occurs with the same time course as that of wild-type amniotic cells—that is, much later and more con-tinuously than the flattening of the serosal cells. Thismay explain the delayed condensation of the embry-onic anlagen after Tc-zen1 RNAi (Figure 3). The slightlywider-spaced nuclei at the anterior tip of early Tc-zen1RNAi embryos might be the first sign of amnion forma-tion (Figure 3G; Figure 4J).

Size Regulation after Loss of Tc-zen1The “big head” phenotype is the most striking conse-quence of a loss of Tc-zen1 and is clearly visiblethroughout germband elongation (Figures 3I and 3J).The head and thorax region of the embryo is enlargedin comparison to the wild-type (Figures 3C, 3D, 3I, and3J). The development of the head and thorax is alsoconsiderably delayed, as indicated by a retarded for-mation of the head lobes and segmental grooves in thisarea. The abdominal segments, however, look normal,probably because they emerge from a growth zone andare unaffected by fate shifts in the blastoderm.

To get a quantitative measure for these changes, weanalyzed the size of the Engrailed (En) stripes (Figures3E and 3K). The third En stripe consisted of 1.5× asmany cells in the lateral dimension in comparison tothe wild-type stripe (n = 4). The total area occupied bythis stripe is even larger than 1.5× the area of a wild-type stripe. Hence, not only does the anterior of the

Current Biology630

embryo consist of more cells, those cells are also less oidensely packed than in the wild-type. The latter may be

the consequence of a slower condensation of head andsthoracic cells.

The enlargement of head and thorax is restored dur- ffing further development, leading to strikingly normal

embryos at the time of dorsal closure (Figure 2; Figures ss3F and 3L). The double membrane covering the abdo-

men is still present (Figure 3L), but it disappears during ttdorsal closure. Finally, 78% of the larvae secreted a

largely normal cuticle: 54% of the larvae even hatched, wfwhereas 24% did not hatch, but their cuticle was indis-

tinguishable from the wild-type except for the head be- rting bent toward the ventral side (Figure 2, second bar;

data not shown). To investigate whether the enlarge- ltment of the head and thorax during early development

led to an increased size of the larvae, we compared the aslength of wild-type and Tc-zen1 knockdown larvae. No

significant difference was detected (Tc-zen1 RNAi: 324 mo�m, n = 6, standard deviation (sd) = 18 �m; wild-type:

335 �m, n = 10, sd = 28 �m). To check whether regula- s[tion of cell number was taking place by decreased cell

divisions or by increased cell death, we performed tαPH3 stainings and TUNEL assays, respectively. How-ever, within the head and thorax region, no apparent f

sdifferences were detected between wild-type and Tc-zen1 knockdown embryos (data not shown). t

tIn summary, these observations show that not onlycan Tribolium embryos develop without a serosa, but l

uthat they also possess the regulative capacity to ame-liorate a severe anterior expansion of the early fate map d

aof the germ rudiment. The mechanisms responsible forthis size regulation are, however, not obvious (see Dis- acussion).

wgTc-zen2 Is Required for Dorsal Closurec

In contrast to Tc-zen1, Tc-zen2 RNAi causes no abnor-c

malities before dorsal closure (Figure 2, third bar). Theembryos form a normal serosa, which expresses Tc-zen1 (data not shown). Of the collected cuticles, how- S

Rever, 57% turned out to be completely everted (Figure2, fourth bar; Figure 6). In these cuticles, the legs, bris- B

dtles, and urogomphi are enclosed by the body wall (Fig-ures 6C and 6D), whereas the tracheae and hindgut lie j

Figure 6. Everted Larval Cuticle after Tc-zen2 RNAi

(A) Wild-type first-instar larva.(B) Everted (inside out) larva after Tc-zen2RNAi.(C and D) Magnified regions from (B). (C) Thelegs are enclosed by the body wall. Arrow-heads mark the tips of the legs. (D) Bristles,marked by arrowheads, and urogomphi,marked by an asterix, point to the inside andare enclosed by the body wall.

utside (data not shown). Wild-type dorsal closure wasnvestigated in order to understand this phenotype.

After closure of the serosal window, the amnion anderosa cover the embryo as separate cell layers. At theully extended germband stage, however, the amnionuses with the serosa beneath the thorax and posterioregments of the head (Figure 7A; see also Figure 8 forchematic representation). Analysis of confocal sec-ions suggests that this fusion takes place by intercala-ion of amniotic and serosal cells. This is in accordanceith the appearance of closely adjacent nuclei in the

usion area, among the otherwise regularly spaced se-osal nuclei (data not shown). The fusion proceeds inhe lateral-to-dorsal direction (Figure 7B), pulling theateral sides of the embryo toward dorsal. At this stage,he embryo is surrounded ventrally only by the fusedmnion and serosa (Figure 7B). This fused amnion anderosa subsequently disappears (Figure 7C). The re-aining serosa, which connects the two lateral sidesf the embryo and covers the yolk at the dorsal side,tarts to crumble (Figure 7C), forming the dorsal organ26]. When the dorsal organ is absorbed into the yolk,he embryo closes dorsally.

After Tc-zen2 RNAi, however, the amnion does notuse with the serosa. Amnion and serosa are visible aseparate cell layers even at the stage of germband re-raction (Figure 7D). Consequently, the dorsal sides ofhe embryo are forced to close ventrally, enclosing theegs and leaving the nerve cord outside (Figure 7E). Fig-res 8I–8N show a schematic comparison of wild-typeorsal closure with the “ventral” closure of the embryofter Tc-zen2 RNAi; the latter type of closure generatesn everted embryo.Because Tc-zen2 is expressed in the late amnion,here fusion of the amnion and the serosa starts, andiven the absence of fusion after Tc-zen2 RNAi, weonclude that Tc-zen2 is necessary for the proper inter-alation of the amnion and the serosa.

imultaneous Loss of Tc-zen1 and Tc-zen2escues the Tc-zen2 Knockdown Phenotypeecause the late amniotic expression of Tc-zen2 is in-ependent of Tc-zen1, we asked whether combined in-

ections of Tc-zen1 and Tc-zen2 dsRNA would result in

Functional Analysis of the Tribolium zen Genes631

Figure 7. Wild-Type Dorsal Closure and the Origin of the Everted Tc-zen2 RNAi Phenotype

Confocal images of cross-sections through the thoracic region of wild-type (A–C) and Tc-zen2 RNAi embryos (D and E). Cell membranes arestained with anti-phospho-tyrosine (red), and cell nuclei are stained with YOYO (green). (A#), (B#), and (D#) show magnifications of the area inthe white squares in (A), (B), and (D), respectively. Schematic drawings are presented to the right.(A) Extended germband stage. The amnion fused with the serosa at the ventral side.(B) Wild-type retracted germband stage. The fusion of amnion and serosa proceeded more toward the dorsal side.(C) Wild-type dorsal closure. The fused amnion and serosa disappeared. The remaining serosa connects the two sides of the embryo abovethe yolk. The asterix marks the dorsal organ.(D) After Tc-zen2 RNAi, the amnion and the serosa stay apart as two separate membranes, even at this retracted germband stage.(E) After Tc-zen2 RNAi, the embryos are completely everted. The nerve cord remains outside. The legs are enclosed by the bodywall. “l”denotes leg, and “n” denotes nerve cord.

an additive phenotype. Surprisingly, combined injec-tions lead to phenotypes indistinguishable from thoseproduced by Tc-zen1 dsRNA injections alone. The em-bryos do not display the Tc-zen2 phenotype and exhibitnormal dorsal closure (Figure 2,fifth and sixth bar). Thisindicates that the only function of Tc-zen2 is the fusionof the amnion with the serosa. Because there is no se-rosa after Tc-zen1 RNAi, the fusion of the amnion andthe serosa is not required. The amnion is oriented in amanner allowing dorsal closure to occur normally (com-pare Figures 8H and 8K).

Discussion

In this study, we show that the two zerknullt homologsin Tribolium have very disctinct, nonredundant func-tions. Tc-zen1 has an early function and specifies theserosal fate at the blastoderm stage. RNAi with Tc-zen1results in loss of serosal fate and expansion of the germrudiment fate, including the amnion. Tc-zen2 has a latefunction and is responsible for the fusion, crucial fordorsal closure, of amnion and serosa at the extendedgermband stage. RNAi with Tc-zen2 results in com-

Current Biology632

Figure 8. Schematic Drawings of the Tc-zenLoss-of-Function Phenotypes

(A–D) Wild-type gastrulation and germbandextension. (E–H) The Tc-zen1 RNAi pheno-type. (I–K) Wild-type dorsal closure. (L–N)The Tc-zen2 RNAi phenotype.(A) The onset of gastrulation can be recog-nized by the formation of the primitive pit. Aclear border forms between the stretched-out cells of the serosa and columnar cells ofthe germ rudiment (dashed line). The dorsalcells of the germ rudiment will form theamnion.(B) Amnion and serosa grow over the poste-rior end of the embryo proper to form theposterior amniotic fold. This occurs to alesser extent at the anterior to form the ante-rior amniotic fold. The crests of the foldsform the serosal window.(C) After closure of the serosal window, theamnion and the serosa are visible as sepa-rate membranes.(D) Germband extension.(E) After Tc-zen1 RNAi, no serosa is present,

and the whole embryo consists of germ rudiment. The presumptive amnion occupies the whole anterior and dorsal side of the egg. Theanlagen of the embryo proper are enlarged.(F) The amniotic cells start to flatten, and the cells of the embryo proper condense.(G) A structure similar to the posterior amniotic fold has formed. The head and thorax of the embryos consist of more cells, which are lessdense, generating the typical “big head” phenotype.(H) When the germband starts to retract, a structure similar to the anterior amniotic fold forms.(I) At the extended germband stage, the amnion and the serosa start to fuse beneath the thorax.(J) The fusion proceeds toward dorsal, pulling the lateral sides of the embryo toward dorsal.(K) The fused amnion and serosa disappears. At the dorsal tip, the remaining serosa starts to crumble, forming the dorsal organ and pullingthe sides of the embryo over the yolk. This finally leads to dorsal closure.(L and M) After Tc-zen2 RNAi, the amnion and serosa do not fuse.(N) The amnion still connects the two lateral sides of the embryo ventrally, forcing the embryo to close ventrally, enclosing the legs, andleaving the nerve cord and yolk outside. The following abbreviations were used: aaf, anterior amniotic fold; paf, posterior amniotic fold; pp,primitive pit; and sw, serosal window.

pletely everted embryos. A schematic interpretation of wthe RNAi phenotypes is presented in Figure 8. a

Although the Tc-zen1 and Tc-zen2 proteins have only s38% amino acid identity and, thus, might bind to dif- sferent cis-regulatory elements, it is more likely that the Adistinct phenotypes are generated by diverged tran- tscriptional regulation. We found that early serosal Tc- tzen2 expression is dependent on Tc-zen1, which ex- uplains why Tc-zen2 does not rescue the serosa after aTc-zen1 RNAi. In addition, Tc-zen1 is not expressed in bthe late amnion, which explains why Tc-zen1 cannot drescue dorsal closure. This makes the duplication of szen in Tribolium a textbook example of subfunctionali- Szation [27, 28]. t

iTExpansion of the Germ RudimentgIn Drosophila, zerknullt mutations lead to cell-fate shifts

along the DV axis as the loss of the amnioserosa isccompensated by an expansion of the dorsal ectodermptoward the dorsal midline. Accordingly, zen expressioncin Drosophila is largely controlled by the DV patterningrsystems, at both the maternal and zygotic level [9, 10].pIn Tribolium, loss of Tc-zen1 causes an enlargementqof the germ rudiment at the expense of the serosa. Thedearly expansions, visualized by molecular markers, oc-scur primarily along the AP axis (Figures 4C–4F and 4K–m4N), suggesting an AP patterning role for Tc-zen1. This

is consistent with the earliest Tc-zen1 expression, t

hich starts symmetrically at the anterior pole withoutdorsal tilt [14, 29]. Furthermore, ventralized and dor-

alized Tribolium embryos possess a dorsoventrallyymmetrical serosa at the anterior pole (P. Kalscheuer,. Basal, and M.v.d.Z., unpublished data). This shows

hat the serosa is primarily established under control ofhe AP patterning system and only later tilts dorsallynder influence of the DV system. The latter is probablydirect regulatory input of the DV pathway on Tc-zen1ecause Tc-zen1 expression slightly expands at theorsal side before the morphological distinction of theerosa and germ rudiment cells becomes apparent.uch a DV input might be the starting point for evolu-

ionary changes that culminate in the situation, foundn Drosophila, in which zen is a DV patterning gene. Inribolium, however, zen1 is primarily an AP patterningene.Tc-zen1 might act like a gap gene in Drosophila. This

ould even account for the enlargement of the moreosterior germ rudiment fates after Tc-zen1 RNAi be-ause deletion of gap genes in Drosophila led to long-ange fate shifts [30]. However, a substantial part of theosterior enlargement is probably a secondary conse-uence of loss of Tc-zen1. Because the germ rudimentoes not condense in the absence of the serosa, thepatial coordinates of the uniform blastoderm stage areaintained until the beginning of gastrulation. The en-

ire blastoderm behaves now like the germ rudiment in

Functional Analysis of the Tribolium zen Genes633

the wild-type, and the cell number increases through-out the embryo surface. This results in true fate-mapshifts along both body axes, as revealed by our analy-sis of the hairy and Engrailed stripes in Tc-zen1 RNAiembryos. Thus, one of the main functions of the serosaprior to gastrulation might be the coordinated scalingdown of an early expanded fate map. Absence of Tc-zen1 maintains the expanded fatemap, generating asituation similar to long-germ development, in whichgenes such as twist are expressed along the entire APaxis.

The shrinkage of an early expanded fate map duringtransition from uniform to differentiated blastoderm isprobably an ancestral feature of short-germ developmentbecause it has been observed by fate-map studies ondragonflies (Odonata), which represent one of the mostprimitive hemimetabolous insect orders [31].

Size RegulationThe dramatic early fate shifts after Tc-zen1 RNAi resultin embryos with an enlarged head and thorax. This en-largement is caused by (1) a lower density and (2) anincreased number of cells contributing to head and tho-rax in comparison to the wild-type. How is it possiblethat normally sized larvae arise from these embryos?

The lower density of cells in the head region iscaused by the absence of flattening serosal cells. Inthe wild-type, the serosa enables the cells of the germrudiment to condense ventrally at the onset of gastrula-tion. Instead of serosal cells, amniotic cells are found atanterior and dorsal positions of Tc-zen1 RNAi embryos.Although the amniotic cells flatten much later and moregradually than the serosal cells, the amniotic cellseventually compensate for the loss of the serosa afterTc-zen1 RNAi and enable the head and thorax to con-dense slowly and normalize the cell density.

It is less clear how the increase in cell number afterTc-zen1 RNAi is corrected. In Drosophila embryos withmultiple copies of bicoid, the head and thorax regionsare enlarged at the expense of the abdominal region[18, 32]. Nevertheless, larvae emerge with an almostnormal cuticle pattern [33]. In this case, it was shownthat local changes in cell division and cell death ratescould account for size regulation [34]. However, neitheranti-PH3 stainings for cell divisions nor TUNEL assaysfor cell death showed obvious differences between Tc-zen1 knockdown and wild-type embryos. Thus, eitherour methods to record changes in cell division or celldeath rates are not sufficiently accurate, or alternativemechanisms of size regulation exist.

One alternative mechanism could be a cell-fate shiftof embryo proper to amnion. Such a mechanism is con-ceivable for two reasons. First, a surplus of amnioticcells does not harm the embryo because an enlargedamnion will just fold further over the abdomen and thehead of the embryo proper (Figure 8H). An excess ofamnioserosa cells does not harm development in Dro-sophila either. In Drosophila embryos with four copiesof decapentapledgic (dpp), the amnioserosa consistsof up to 400 cells instead of the usual 135, with no com-promise in viability [35]. Second, the fate decision be-tween the amnion and the embryo proper is probablymore flexible than the fate decision between the serosa

and the germ rudiment. In ligation experiments with thecamel cricket Tachycines, for example, amniotic cellscan be induced to form missing parts of the germband[36, 37]. Therefore, a progressive transformation of em-bryonic into amniotic tissue may contribute to the sizeregulation of the enlarged head and thorax in Tc-zen1RNAi embryos.

Dorsal ClosureThe classical description of dorsal closure in short-germ insects assumes a reversal of the movements thathave generated amnion and serosa in the first place.The serosa fuses with the amnion, and a serosal win-dow forms that widens and finally releases the embryoat the ventral side. However, this course of events isfound only in ancestral hemimetabolous insect orders[38]. In the majority of cases, the serosa stays con-nected to the inner eggshell or the serosal cuticle,whereas dorsal closure appears to be largely driven bymorphogenetic movements within the amnion or dorsalectoderm of the embryo [5, 38].

We have seen a similar phenomenon in Tribolium(Figures 8I–8K). The fusion of amnion and serosa ven-trally does not lead to a renewed formation of a serosalwindow. The amnion appears to intercalate with the se-rosa, dragging the two halves of the embryonic ecto-derm toward the dorsal side of the embryo. A functionin dorsal closure is probably ancestral for zen becausezen is also expressed in the late amnion in Schistocerca[15]. More importantly, RNAi with zen in the hemipteranOncopeltus fasciatus results in everted embryos (K.A.Panfilio, P.Z. Liu, and T.K. Kaufman, personal communi-cation).

The function of Tc-zen2 is restricted to mediating thefusion between amnion and serosa. After serosa lossdue to Tc-zen1 RNAi, Tc-zen2 is not required anymorefor dorsal closure. The remaining amnion then connectsthe two sides of the embryo dorsally and covers theyolk, enabling normal dorsal closure (compare Figures8H and 8K). The situation resembles normal develop-ment in Apis or Drosophila. In Apis, the amnion doesnot enclose the embryo by forming an amniotic cavity,but covers the yolk at the dorsal side [39]. In Drosoph-ila, a single extraembryonic cell layer, the amnioserosa,covers the yolk and embryo dorsally and is indispens-able for dorsal closure.

Taken together, the loss of the serosa in the Triboliumembryo reveals an unexpected plasticity of the extra-embryonic membrane system in a short-germ insect. Asituation is generated that shows similarity to normaldevelopment of more-derived long-germ embryos fromother holometabolous insect orders (Hymenoptera andDiptera). The plasticity seen in Tribolium might repre-sent the ancestral condition and might have facilitatedthe evolutionary changes that led to the reduction ofthe extraembryonic membranes in hymenopterans andhigher dipterans.

The Function of the SerosaNormally sized larvae and perfect dorsal closure afterthe loss of the serosa leaves us with a baffling problem:Why do Tribolium embryos need a serosa in the firstplace? Protection against desiccation or rupture [1] is

Current Biology634

not essential because the Tc-zen1 RNAi embryos with- eaout serosa survived the dry laboratory conditions and

relatively rough handling with sieves. Nevertheless, we plcan exclude neither an essential physical protection

against conditions that are found only in the natural cchabitat nor a small contribution, which would still be

highly significant under conditions of natural selection, eto the survival rate due to physical protection. However,

Ean interesting alternative to physical protection has re-cently been suggested. In Manduca, extraembryonic

Stissues appear to harbor innate immune functions that Bprotect the embryo against bacterial infections [40]. In mTribolium, the NF-κB protein Dorsal is expressed at c

mhigh levels in the serosa [29]. Because the Toll-rel/NF-mκB pathway is crucial for innate immunity [41, 42], an(immune function had been suggested for the serosa ins

Tribolium as well [29]. Such a function would be appar-ent only if the embryos were challenged by pathogens. E

EfHox3 and BicoidEIt has been assumed that zen is derived from an an-qcestral class-3 Hox gene [43]. In more primitive arthro-l

pods, the Hox3/zen homolog is expressed in a ca- enonical Hox3-like fashion, as shown for the spider tCupiennius salei and the mite Archegozetes longiseto-

Csus [44, 45]. The primitive wingless insect ThermobiaRprovides an interesting intermediate case [46]. There,mzen is expressed as a Hox3 gene in head segments, butT

also in the emerging amnion, where it possibly fulfils a Cmorphogenetic role. In the lineage leading to the Awinged insects, Hox3 completely lost its function in 6specifying segment identity and expanded its role in

Sthe morphogenesis and specification of the extraem-dbryonic tissues. Concomitantly, it became involved ina

specifying the serosa that arises from anterior egg re- pgions in the majority of insects. The anterior expression Drequired for this function of zen might be due to a new w

aenhancer element. However, it might also be derivedffrom an ancestral enhancer element that still had bind-Ting sites for AP patterning genes. In the latter case, theG

AP patterning function of Tc-zen1 would be a vestige iof its homeotic origin. T

Irrespective of the evolutionary path by which zen ac- squired a function in serosa specification, this event was

Pprobably crucial for the evolution of the anterior mater-Fnal determinant bicoid, which is involved in head andsthorax formation in higher dipterans [18, 47–49]. In-w

deed, it is striking that there is a formal similarity be- dtween the phenotypes caused by loss of bicoid in Dro- asophila and those caused by loss of zen1 in Tribolium. w

aIn both cases, anterior regions of the early embryo aredeleted and replaced by the expansion of more poste-

Irior regions. In a long-germ insect such as Drosophila,I

the anterior regions correspond to the anlagen of head pand thorax; in a short-germ insect such as Tribolium, sthey correspond to the anlagen of the serosa. However, d

tthe serosa is not only the anteriormost fate of the blas-toderm in most insects—the decision between serosa

Cand germ rudiment is also one of the earliest cell-dif-Eferentiation events. Thus, zen had to acquire an earlyr

anterior expression. This requirement has probably fa- mvored evolutionary changes that allowed early tran- w

cscription and anterior localization, including maternal

xpression. Both Schistocerca zen [15] and Tc-zen1 arelready maternally expressed (S. Brown and L. Farzana,ersonal communication; M.v.d.Z. and S.R., unpub-

ished data). Taken together, these data make it con-eivable that the involvement of zen in serosa specifi-ation provided a favorable starting point for thevolution of bicoid in higher dipterans [47–49].

xperimental Procedures

tock Maintenanceeetles were kept on wheat flour (Diamant extra, type 405) supple-ented with dried baker’s yeast in plastic boxes at 30°C. For egg

ollection, beetles were placed for 1–3 days on instant flour (Dia-ant instant, type 405). Beetles were sieved out with a 710 �maze sieve, and eggs were collected with a 300 �m maze sieve

Retsch). All food was kept at least one night at −20°C, and allieves were kept at 60°C in order to prevent infections.

mbryo Fixationmbryos were dechorionized with a hypochlorite solution and fixed

or 30 min in 4 ml PEMS (0.1 M PIPES, 2 mM MgSO4, and 1 mMGTA [pH 6.9]), 5 ml heptane, and 1 ml 20% formaldehyde. Subse-uently, the water phase was removed, and embryos were devitel-

inized by a methanol shock and stored in methanol at −20°C. Oldermbryos, difficult to devitellinize by a methanol shock, were devi-ellinized with forceps and needle.

loning of Tc-zen2NA was isolated by a trizol (Invitrogen) extraction, and cDNA wasade with the Cloned AMV First Strand Synthesis kit (Invitrogen).

c-zen2 was amplified from this cDNA with the forward primer 5#-CATTCTCGGGGCTTTTCATAG-3# and the reverse primer 5#-ACATTCTTCCCTTGGTAATACTG-3# at an annealing temperature of0°C and cloned into TOPO vector (Invitrogen).

ynthesis of dsRNA and In Situ ProbessRNA was synthesized with the MEGAscript RNAi kit of Ambionccording to the manufacturer’s protocol. DIG-labeled in siturobes were made with the MAXIscript T7/T3 Kit (Ambion), but withIG RNA Labeling mix (Roche). Preferentially linearized plasmidsere used as template instead of PCR products. For lacZ dsRNA,fragment including the lacZ gene and a T7 site was amplified

rom the TOPO II vector with the forward primer 5#-AGCGCCCAAACGCAAACCG-3# and the reverse primer 5#-CACACCCGCCCGCTTAATG-3# at an annealing temperature of 65°C and cloned

nto the TOPO 2.1 vector (Invitrogen). Fragments with two opposing7 sites were excised with PvuII, purified (Purification kit, Amer-ham), and used as template for dsRNA synthesis.

arental RNAiemale pupae were collected shortly before hatching, and the dor-al side of the terminal segment was fixed on a microscope slideith Fixogum (Marabu). Approximately 0.2 �l of a 0.5–1.0 �g/�lsRNA solution was ventrally injected between the third and fourthbdominal segment (see Figure 2 for total amounts). After 5 days,ild-type male beetles were added, and offspring were collectednd analyzed (after [50]).

n Situ Hybridizations and Immunostainingsn situ hybridizations were performed as described [51], but withoutroteinase-K treatment. In situ stained embryos were counter-tained with DAPI. Immunostainings were carried out essentially asescribed in [52]. The Engrailed antibody 4D9 was used 1:5, and

he anti-phospho-histone3 antibody was used 1:1000 in PBST.

uticle Preparationggs were transferred to a 96 well plate. After 5 days, hatching

ates were counted, and a few drops of a 9:1 lactic acid:ethanolixture were added to every well. After one night at 60°C, cuticlesere studied under a dark-field binocular or mounted for light mi-roscopy.

Functional Analysis of the Tribolium zen Genes635

TUNEL AssaysAssays were performed essentially as described in [53]. However,0.1% Tween-20 was added to the TdT and DNase I buffer.

Preparation of Cross-Sections for Confocal MicroscopyEmbryos in their vitelline membrane were cut with a razor bladeinto five or six slices in PBST, treated with RNAse, and washed inPBST. After being refixed for 10 min in 4% formaldehyde in PBST,sections were washed with PBST, blocked for 1 hr in 1% bovineserum albumin and 3% normal goat serum in PBST, and subse-quently incubated with undiluted anti-phospho-tyrosine antibodyovernight at 4°C. After being washed with PBST and blocked for30 min, sections were incubated for 2 hr with Alexa 555 anti-mouseantibody (Molecular Probes, 1:400) and YOYO-1 (Molecular Probes,1:25.000) in PBST at room temperature. After being washed withPBST, sections were embedded in vectashield (Vector Laborato-ries) and examined under a confocal microscope.

Acknowledgments

We thank Magdalena Baer for doing the first Tc-zen2 in situs, JoëlSavard for the generous supply of marker genes from his EST col-lection, and Reinhard Schröder for the gift of Tc-otd1. Niko Prpichelped with the TUNEL assay. We are grateful to Claude Desplan,Diethard Tautz, and Wim Damen for various comments on themanuscript. John Baines improved the English. M.v.d.Z. thanks Ab-idin Basal for the introduction into Tribolium and Patrick Kalscheuerand other lab members for advice. M.v.d.Z. was supported by theInternational Graduate School in Genetics and Functional Geno-mics of the University of Cologne.

Received: December 15, 2004Revised: February 21, 2005Accepted: February 21, 2005Published: April 12, 2005

References

1. Zeh, D.W., Zeh, J.A., and Smith, R.L. (1989). Ovipositor, amni-ons and egg shell architecture in the diversification of ter-restrial arthropods. Q. Rev. Biol. 64, 147–168.

2. Machida, R., and Ando, H. (1998). Evolutionary changes in de-velopmental potentials of the embryo proper and embryonicmembranes along with the derivative structures in atelocerata,with special reference to the hexapoda (arthropoda). Proc. Ar-thropod. Embryol. Soc. Jpn. 33, 1–13.

3. Roth, S. (2004). Gastrulation in other insects. In Gastrulation:From Cells to Embryos, C. Stern, ed. (New York: Cold SpringHarbor Laboratory Press), pp. 105–121.

4. Anderson, D.T. (1972a). The development of hemimetabolousinsects. In Developmental Systems: Insects, S.J. Counce andC.H. Waddington, eds. (London: Academic Press), pp. 96–163.

5. Schwalm, F.E. (1988). Insect Morphogenesis (Basel: Karger).6. Schmidt-Ott, U. (2000). The amnioserosa is an apomorphic

character of cyclorrhaphan flies. Dev. Genes Evol. 210, 373–376.

7. Handel, K., Grunfelder, C.G., Roth, S., and Sander, K. (2000).Tribolium embryogenesis: A SEM study of cell shapes andmovements from blastoderm to serosal closure. Dev. GenesEvol. 210, 167–179.

8. Rushlow, C., and Levine, M. (1990). Role of the zerknullt genein dorsal-ventral pattern formation in Drosophila. Adv. Genet.27, 277–307.

9. Rushlow, C., Frasch, M., Doyle, H., and Levine, M. (1987). Ma-ternal regulation of zerknullt: A homoeobox gene controllingdifferentiation of dorsal tissues in Drosophila. Nature 330,583–586.

10. Raftery, L.A., and Sutherland, D.J. (2003). Gradients andthresholds: BMP response gradients unveiled in Drosophilaembryos. Trends Genet. 19, 701–708.

11. Wakimoto, B.T., Turner, F.R., and Kaufman, T.C. (1984). Defectsin embryogenesis in mutants associated with the Anten-

napedia Gene Complex of Drosophila melanogaster. Dev. Biol.102, 147–172.

12. Rushlow, C., Doyle, H., Hoey, T., and Levine, M. (1987). Molecu-lar characterization of the zerknullt region of the Antennapediagene complex in Drosophila. Genes Dev. 1, 1268–1279.

13. Pultz, M.A., Diederich, R.J., Cribbs, D.L., and Kaufman, T.C.(1988). The proboscipedia locus of the Antennapedia Complex:A molecular and genetic analysis. Genes Dev. 2, 901–920.

14. Falciani, F., Hausdorf, B., Schröder, R., Akam, M., Tautz, D.,Denell, R., and Brown, S. (1996). Class 3 Hox genes in insectsand the origin of zen. Proc. Natl. Acad. Sci. USA 93, 8479–8484.

15. Dearden, P., Grbic, M., Falciani, F., and Akam, M. (2000). Mater-nal expression and early zygotic regulation of the Hox3/zengene in the grasshopper Schistocerca gregaria. Evol. Dev. 2,261–270.

16. Brown, S.J., Fellers, J.P., Shippy, T.D., Richardson, E.A., Max-well, M., Stuart, J.J., and Denell, R.E. (2002). Sequence of theTribolium castaneum Homeotic Complex: The region corre-sponding to the Drosophila melanogaster Antennapedia Com-plex. Genetics 160, 1067–1074.

17. Randazzo, F.M., Seeger, M.A., Huss, C.A., Sweeney, M.A., Ce-cil, J.K., and Kaufman, T.C. (1993). Structural changes in theAntennapedia Complex of Drosophila pseudoobscura. Genet-ics 133, 319–330.

18. St Johnston, D., and Nusslein-Volhard, C. (1992). The origin ofpattern and polarity in the Drosophila embryo. Cell 68, 201–219.

19. Sommer, R.J., and Tautz, D. (1994). Expression pattern of twistand snail in Tribolium (Coleoptera) suggest a homologous for-mation of mesoderm in long and short-germ band insects. Dev.Genet. 15, 32–37.

20. Handel, K., Basal, A., Fan, X., and Roth, S. (2005). Triboliumcastaneum twi: Gastrulation and mesoderm formation in ashort germ beetle. Dev. Genes Evol. 215, 13–31.

21. Savard, J. (2004). A Genomic Approach to the Study ofTribolium castaneum Genetics, Development and Evolution.PhD thesis, University of Cologne, Cologne, Germany.

22. Li, Y., Brown, S.J., Hausdorf, B., Tautz, D., Denell, R.E., andFinkelstein, R. (1996). Two orthodenticle-related genes in theshort-germ beetle Tribolium castaneum. Dev. Genes Evol. 206,35–45.

23. Brown, S.J., Parrish, J.K., Beemann, R.W., and Denell, R.E.(1997). Molecular characterization and embryonic expressionof the even-skipped ortholog of Tribolium castaneum. Mech.Dev. 61, 165–173.

24. Sommer, R.J., and Tautz, D. (1993). Involvement of an or-thologue of the Drosophila pair-rule gene hairy in segment for-mation of the short germ-band embryo of Tribolium (Coleop-tera). Nature 361, 448–450.

25. Berns, N. (2001). Untersuchungen zur Struktur und Funktiondes Tribolium Brachyury-Homologs Tcbyn. PhD thesis, Univer-sity of Tubingen, Tubingen, Germany.

26. Anderson, D.T. (1972). The development of holometabolous in-sects. In Developmental Systems: Insects, Volume 1, S.J.Counce and C.H. Waddington, eds. (London: Academic Press),pp. 165–271.

27. Force, A., Lynch, M., Pickett, F.B., Amores, A., Yan, Y., andPostlethwait, J. (1999). Preservation of duplicate genes bycomplementary, degenerative mutations. Genetics 151, 1531–1545.

28. Lynch, M., and Force, A. (2001). The probability of duplicategene preservation by subfunctionalization. Genetics 154, 459–473.

29. Chen, G., Handel, K., and Roth, S. (2000). The maternal NF-κB/Dorsal gradient of Tribolium castaneum: Dynamics of earlydorsoventral patterning in a short-germ beetle. Development127, 5145–5156.

30. Rivera-Pomar, R., and Jäckle, H. (1996). From gradients tostripes in Drosophila embryogenesis: Filling the gaps. TrendsGenet. 12, 478–483.

31. Seidel, F. (1935). Der Anlageplan im Libellenei, zugleich eineUntersuchung uber die allgemeinen Bedingungen fur defekteEntwicklung und Regulation bei dotterreichen Eiern. RouxsArch. Entw. Mech. Org. 132, 671–751.

Current Biology636

32. Frohnhöfer, H.G., and Nusslein-Volhard, C. (1986). Organizationof anterior pattern in the Drosophila embryo by the maternalgene bicoid. Nature 324, 120–125.

33. Busturia, A., and Lawrence, P. (1994). Regulation of cell numberin Drosophila. Nature 370, 561–563.

34. Namba, R., Pazdera, T.M., Cerrone, R.L., and Minden, J.S.(1997). Drosophila embryonic pattern repair: How embryos re-spond to bicoid dosage alteration. Development 124, 1391–1403.

35. Wharton, K.A., Ray, R.P., and Gelbart, W.M. (1993). An activitygradient of decapentapledgic is necessary for the specificationof dorsal pattern elements in the Drosophila embryo. Develop-ment 117, 807–822.

36. Krause, G. (1952). Schnittoperation im Insekten-Ei zum Nach-weis komplementärer Induktion bei Zwillingsbildung. Naturwis-senschaften 39, 356.

37. Sander, K. (1976). Specification of the basic body pattern ininsect embryogenesis. Adv. Insect Physiol. 12, 125–238.

38. Sander, K. (1976). Morphogenetic movements in insect embry-ogenesis. In Insect Development, P.A. Lawrence, ed. (Oxford:Blackwell), pp. 35–52.

39. Fleig, R., and Sander, K. (1988). Honeybee morphogenesis:Embryonic cell movements that shape the larval body. Devel-opment 103, 525–534.

40. Gorman, M.J., Kankanala, P., and Kanost, M.R. (2004). Bacte-rial challenge stimulates innate immune responses in extra-embryonic tissues of tobacco hornworm eggs. Insect Mol. Biol.13, 19–24.

41. Anderson, K.V. (2000). Toll signalling pathways in the innate im-mune response. Curr. Opin. Immunol. 12, 13–19.

42. Hoffmann, J.A., Kafatos, F.C., Janeway, C.A., and Ezekowitz,R.A. (1999). Phylogenetic perspective in innate immunity. Sci-ence 284, 1313–1318.

43. Brown, S., Fellers, J., Shippy, T., Denell, R., Stauber, M., andSchmidt-Ott, U. (2001). A strategy for mapping bicoid on thephylogenetic tree. Curr. Biol. 11, R43–R44.

44. Damen, W.G.M., and Tautz, D. (1998). A Hox class 3 orthologuefrom the spider Cupiennius salei is expressed in a Hox-gene-like fashion. Dev. Genes Evol. 208, 586–590.

45. Telford, M.J., and Thomas, R.H. (1998). Of mites and zen: Ex-pression studies in a chelicerate arthropod confirm zen is adivergent Hox gene. Dev. Genes Evol. 208, 591–594.

46. Hughes, C.L., Liu, P.Z., and Kaufman, T.C. (2004). Expressionpatterns of the rogue Hox genes Hox3/zen and fushi tarazuin the apterygote insect Thermobia domestica. Evol. Dev. 6,393–401.

47. Stauber, M., Jaeckle, H., and Schmidt-Ott, U. (1999). The ante-rior determinant bicoid of Drosophila is a derived Hox class 3gene. Proc. Natl. Acad. Sci. USA 96, 3786–3789.

48. Stauber, M., Prell, A., and Schmidt-Ott, U. (2002). A single Hox3gene with composite bicoid and zerknullt expression charac-teristics in non-Cyclorrhaphan flies. Proc. Natl. Acad. Sci. USA99, 274–279.

49. Stauber, M., Taubert, H., and Schmidt-Ott, U. (2000). Functionof bicoid and hunchback homologs in the basal cyclorrhaphanfly Megaselia. Proc. Natl. Acad. Sci. USA 97, 10844–10849.

50. Bucher, G., Scholten, J., and Klingler, M. (2002). Parental RNAiin Tribolium (Coleoptera). Curr. Biol. 12, R85–R86.

51. Tautz, D., and Pfeifle, C. (1989). A non-radioactive in situ hy-bridization method for the localization of specific RNAs in Dro-sophila embryos reveals translational control of the segmenta-tion gene hunchback. Chromosoma 98, 81–85.

52. Roth, S., Stein, D., and Nusslein-Vollhard, C. (1989). A gradientof nuclear localisation of the dorsal protein determines dorso-ventral pattern in the Drosophila embryo. Cell 59, 1189–1202.

53. Prpic, N.-M. (2004). Vergleichende Studien zur Gliedmaßenent-wicklung bei Arthropoden. PhD thesis, University of Cologne,Cologne, Germany.