study of the posterior spiracles of drosophila as a model ... · model to understand the genetic...

csrpcfitcdw

Developmental Biology 214, 197–210 (1999)Article ID dbio.1999.9391, available online at http://www.idealibrary.com on

Study of the Posterior Spiracles of Drosophila as aModel to Understand the Genetic and CellularMechanisms Controlling Morphogenesis

Nan Hu and James Castelli-GairDepartment of Zoology, University of Cambridge, Downing Street,Cambridge CB2 3EJ, United Kingdom

We have studied the posterior spiracles of Drosophila as a model to link patterning genes and morphogenesis. A geneticascade of transcription factors downstream of the Hox gene Abdominal-B subdivides the primordia of the posteriorpiracles into two cell populations that develop using two different morphogenetic mechanisms. The inner cells that giveise to the spiracular chamber invaginate by elongating into “bottle-shaped” cells. The surrounding cells give rise to arotruding stigmatophore by changing their relative positions in a process similar to convergent extension. The geneticascades regulating spiracular chamber, stigmatophore, and trachea morphogenesis are different but coordinated to form aunctional tracheal system. In the posterior spiracle, this coordination involves the control of the initiation of cellnvagination that starts in the cells closer to the trachea primordium and spreads posteriorly. As a result, the opening of theracheal system shifts back from the spiracular branch of the trachea into the posterior spiracle cells. We analyze theontribution of the ems gene to this coordination. In ems mutants, invagination of the spiracle cells adjacent to the tracheaoes not occur, but more posterior cells of the spiracle invaginate normally. This results in a spiracle without a lumen andith the tracheal opening located outside it. © 1999 Academic Press

Key Words: tracheal system; morphogenesis; Abd-B; bottle cells; convergent extension.

mttp

ttScTtihtsm1ho

INTRODUCTION

Many organisms are subdivided into segments, each ofthem having specific structures at defined positions.Segment-specific structures are formed because the activa-tion of patterning genes by a system of dorsoventral andanteroposterior information results in the localized activa-tion of specific morphogenetic programs. The best knowngenetic patterning system is the one laid down by the HOXtranscription factors. The information encoded by the Hoxgenes has to be translated into the construction of morpho-logical structures by the control of downstream targets(Botas, 1993; Graba et al., 1997). This means that Hox genesselect the development of a structure by locally controlling,through their targets, the morphogenetic behavior of naivecells. The final morphology of a structure will thereforedepend on the combination of morphogenetic mechanismsthat are activated during development in the cells that formit. The different morphogenetic mechanisms that can be
activated include, among others, the control of cell migra-tion, shape, division, death, and adhesion. The regulation of
si

0012-1606/99 $30.00Copyright © 1999 by Academic PressAll rights of reproduction in any form reserved.

orphogenesis has an added element of complexity due tohe fact that the formation of a structure cannot be con-rolled in isolation because functionality depends on itsroper connection with other elements of the system.To understand the morphogenesis of a structure at least

hree factors have to be taken into account. First, which arehe genes specifying the cells that compose the organ?econd, what are the morphogenetic mechanisms that theells forming the organ use to acquire the final shape?hird, how does the organ connect with other elements of

he system to form a functional unit? Due to the complex-ty of understanding morphogenesis at all three levels, weave decided to study the formation of a relatively simplehree-dimensional organ: the posterior spiracle of the Dro-ophila larva. The spiracles and the trachea connect toake a functional tracheal system (Manning and Krasnow,

993) and they can be considered as a model to understandow genes are involved in the generation of integratedrgans. Spiracles are the openings of the insect tracheal
ystem. Externally they connect to the epidermis andnternally to the tracheal trunk. The spiracles and the
197

a

pvmm2

osAwof

aovaal(clolpdsptt

esC

1ip1pp7

198 Hu and Castelli-Gair

trachea form from separate populations of cells. The tracheaforms from 10 tracheal pits, 1 per hemisegment. Thetracheal pits are ectodermal invaginations that branch andfuse together forming the tracheal network. The trachea isattached to the epidermis by the spiracular branch. Theattachment of the spiracular branch and the epidermisoccurs at the place where the tracheal pit originally invagi-nated. The spiracular branch lumen, however, collapses insegments where no spiracles form and does not participatein respiration. The only function of these collapsed spiracu-lar branches is to shed the tracheal cuticle after each molt(Manning and Krasnow, 1993). The Drosophila larva hasonly two pairs of functional spiracles, the anterior in thefirst thoracic segment (T1) and the posterior in the eighthabdominal segment (A8). The spiracles are formed fromectodermal cells adjacent to the first and last tracheal pits.During development the spiracles connect to the T1 and A8spiracular branches, forming a continuous lumen. Theanterior spiracle is not functional until late larval stages;therefore, all the gas exchange in the young larva occursthrough the posterior spiracle (Manning and Krasnow,1993).

In this paper we study the morphogenesis of the posteriorspiracle. We show that the formation of the posteriorspiracle is achieved by two main morphogenetic mecha-nisms: The internal cells change their shape, whereas theexternal cells rearrange by intercalation. These cellularmechanisms are controlled by a cascade of transcriptionfactors downstream of the Abdominal-B (Abd-B) gene thatpattern the posterior spiracle when it is still a two-dimensional structure. We also show that the mechanismconnecting the posterior spiracle to the trachea differs fromthat used to connect the segmental portions of the tracheainto a single network.

MATERIALS AND METHODS

Antibodies and enhancer trap lines used as markers. We haveused the following primary antibodies: rabbit anti-spalt (Kuhnleinet al., 1994), anti-cut (Blochinger et al., 1990), anti-klumpfuss (Yanget al., 1997), anti-pdm1 (pdm-1 is a synonym of nubbin) (Yeo et al.,1995), anti-b-galactosidase (Cappel); mouse anti-engrailed (Patel etl., 1989), anti-crumbs (Tepass et al., 1990), anti-Abd-B (Celniker et

al., 1989), and anti-b-galactosidase (Promega). The following en-hancer trap lines were used trh3, P837 (A1-3-10), P927 (B7-2-22),P1028 (A495.1M2), P1120 (A189.2F3), and btl-lacZ6.81 (Bellen etal., 1989; Bier et al., 1989; Hartenstein and Jan, 1992; Isaac andAndrew, 1996; Klambt et al., 1992). We have also used as markersconstructs in which the expression of b-galatosidase was driven byspecific enhancers, these are: grh-D4 (a gift from Sarah Bray),ems-1.2 (Jones and McGinnis, 1993); and ct-D2.3 and ct-A4.2 (Jackand DeLotto, 1995).

GAL4 and UAS stocks. GAL4 and UAS stocks included: klu-GAL4 (Klein and Campos-Ortega, 1997), Arm-GAL4 (Sanson et al.,1996), UAS-dpp (Staehling-Hampton and Hoffmann, 1994), UAS-
lacz (cytoplasmic) (Brand and Perrimon, 1993), and UAS-Nod-lacZ(minus end microtubule motor) (Clark et al., 1997). p
Copyright © 1999 by Academic Press. All right

Cuticle preparations and antibody stainings. For cuticlereparations, 24-h-old embryos were dechorionated in bleach. Theitelline membrane was removed by treatment with heptane andethanol. After washing the embryos in 0.1% Tween they wereounted in Hoyer’s mountant and kept on a hot plate at 60°C fordays.For antibody stainings all primary antibodies were incubated

vernight in BBT buffer (Leiss et al., 1988), adding BSA and Goaterum, washed in PBS, and developed using the Vectastain EliteBC kit. For double stainings nickel chloride salts were addedhen developing the first primary antibody, but not when devel-ping the second primary, to allow us to distinguish cells positiveor each antibody.

Nuclei counts. Embryos were stained with anti-sal antibodiesnd the number of nuclei in three places scored. The scoring wasnly performed in the spiracle that was well oriented toward theiewer. Ten or more spiracles where scored for stages 12, 13, and 14nd 7 spiracles for stage 15. The numbers recorded for each spiraclere the circle of sal-expressing cells further away from the spiracu-ar chamber, which is represented as a blue line in Figs. 7B9 and 7C9the basal circumference of the stigmatophore); the circle of cells inontact with the spiracular chamber, which is represented as a redine in Figs. 7B9 and 7C9 (top of the stigmatophore); and the widthf the circle of sal-expressing cells, which is represented as a greenine in Figs. 7B9 and 7C9 (will become the height of the stigmato-hore). The latter two counts are more accurate than the first oneue to the fact that after st14, the base of the left and righttigmatophores fuse and the decision as to where each stigmato-hore ends is a subjective matter. The segmental groove allows uso distinguish unanbiguously the sal expressing cells in A8 fromhose in A9.

Clones. The Tau-bgal clones were induced by MarcosGonzalez-Gaitan, who allowed us to use them for this study. Theclones were induced using an FRT flip-out cassette made by A.Martı-Subirana and R. Holmgren (unpublished).

Mutant fly strains. The following mutant alleles were used:Abd-BM1 (Casanova et al., 1986), ctdb7 (Blochlinger et al., 1988),ms9H83, ems9Q64 (Dalton et al., 1989), trh8 (Isaac and Andrew, 1996),alIIB57 (Kuhnlein et al., 1994), Df(2L) 5 deficient for sal and sal-r (deelis et al., 1996), btlLG19 (Klambt et al., 1992), Df(2)GR4 (deficient

for pdm-1 (5nub) and the structurally related gene pdm-2) (Yeo etal., 1995), Df(3L) kluXR19 (Klein and Campos-Ortega, 1997), grhB32

(Bray and Kafatos, 1991), Df(2L) enE (deficient for en and invected)(Tabata et al., 1995), and Df(3L)H99 (deficient for the three celldeath genes reaper, grim, and head involution defective) (Chen etal., 1996; Grether et al., 1995; White et al., 1994).

RESULTS

The posterior spiracles are formed in the dorsal part of theeighth abdominal segment (Fig. 1). At 6 h of development (st1) the morphology of the cells that will form the spiraclesn A8 is indistinguishable from that of cells at homologousositions in more anterior abdominal segments (Figs. 1E–H). At 13 h (st16), the posterior spiracle has all the featuresresent in the mature spiracle. Therefore, the basic mor-hogenesis of the posterior spiracles is completed in abouth.
The posterior spiracle is an ectodermal structure com-
osed of two parts: the spiracular chamber and the stig-

s of reproduction in any form reserved.

C1esAetc

sstiAacti1tieopaabeen double stained with anti-Ubx (brown), which is not expressedin the dorsal epidermis of A8. Scale bars, 10 mm.

199Posterior Spiracle Morphogenesis in Drosophila


matophore (Fig. 1D). The spiracular chamber is the internaltube connecting the trachea to the exterior. In the larva thistube forms a very refractile filter, the filzkorper (Fig. 1A).The opening of the spiracular chamber, the stigma, issurrounded by four sensory organs: the spiracular hairs.Clones labeling the spiracular hairs show that each one isformed by four cells related by lineage, two neural and twosupport cells, the typical structure of a type I externalsensory organ (data not shown; Jan and Jan, 1993). Thestigmatophore is an external protrusion in which the spi-racular chamber is located. When the larva is buried in thesemiliquid medium where it feeds, the stigmatophore peri-scopes out of the medium, allowing the larva to continuebreathing.

Although functionally related, the stigmatophore, thespiracular chamber, and the trachea develop in differentways. This is reflected in the timing of their developmentand by the genes required during development.

Genes Expressed and Required for PosteriorSpiracle Development

To understand the genetic mechanisms involved in pos-terior spiracle morphogenesis we have started by findingout the regulatory network required for its specification.There are a number of known genes that when mutantaffect the posterior spiracle. The formation of the posteriorspiracles in A8 requires the HOX transcription factorABD-B (Sanchez-Herrero et al., 1985); however, ABD-Bexpression is not restricted to the posterior spiracles but isexpressed in the whole of A8 and in other segments (Celni-ker et al., 1989). The earliest differentiation of the spiraclecells can be detected at the extended germ band stage (6 h ofdevelopment) when a small number of transcription factorsare activated in the dorsal region of A8 (Figs. 1E and 1H).These are later followed by other genes activated in subsetsof cells of the posterior spiracles.

To analyze the regulatory relations between these genes,we have studied how the expression pattern of each one isaffected in mutants for the others (Table 1). These resultsshow that the transcription factors expressed in the spiraclecan be organized in a hierarchical cascade (Fig. 2). Theexpression of all the genes studied requires the function ofABD-B as in mutants for Abd-B no spiracle specific geneexpression is detected in A8.

Downstream of Abd-B, the cascade can be subdividedinto various levels. The activation of cut (ct), empty spi-racles (ems), nubbin (nub), klumpfuss (klu), and spalt (sal)genes (Blochinger et al., 1990; Dalton et al., 1989; Klein and

ampos-Ortega, 1997; Kuhnlein et al., 1994; Ng et al.,995; Walldorf and Gehring, 1992) does not require thexpression of any of the other genes that we have studied,uggesting that they are at the top of the cascade underBD-B regulation. The ct, ems, nub, and klu genes are

xpressed in the spiracular chamber in overlapping pat-

FIG. 1. Posterior spiracle structure at early and late stages ofdevelopment. Three-dimensional structure of the posterior spi-racles after 15 h of development (left) and two-dimensional orga-nization of the precursor cells of the posterior spiracles at 6 h ofdevelopment (right). Posterior spiracle of a newly eclosed larva (A)and schematic section of a spiracle (D). In all figures dorsal is upand anterior to the left. The trachea (tr.) connects to the spiracularchamber (sp. ch.) where the filzkorper will form. The filzkorper(Fz.) is a refractile filter formed by cuticle extensions of the cells ofthe spiracular chamber. The spiracular chamber opens to theoutside through the stigma around which there are four branchedspiracular hairs (h.). Single cell clones marked with Tau–b-gal (B)howing the elongated “bottle” cell shape of the invaginatedpiracular chamber cells. In (C) two of the support cells giving riseo a spiracular hair are marked (black arrowhead), as well as a singlenternal bottle shape cell (white arrowhead). Close up of segments6–A8 at stage 11 stained with anti ct (E–G) or anti-nub (H). Bothntibodies mark in A8 the cells that will give rise to the spiracularhamber. The embryo in G has also been stained with anti-en andhe embryo in F with anti-b-gal to detect the expression of an insertn the breathless gene that labels the tracheal pits. Note that at st1 (6 h) the cells of the spiracular chamber can only be identified byhe expression of the antibodies, as they are morphologicallyndistinguishable from cells in more anterior segments. The ct-xpressing cells (black staining in E–G) are ectodermal cells locatedn the surface of the embryo, arranged as a two-dimensional sheetosterior to the opening of the tracheal pit. Double staining withnti-en (G) reveals that all ct-expressing cells (black) are in thenterior compartment of the A8 segment. The embryo in (H) has

erns. The sal gene is not expressed in the spiracularhamber but in the cells that surround it and will form the


ss

ah

Kaerws

educe

to


stigmatophore. The exclusion of sal from the spiracularchamber is partly due to repression by ct, as in ct mutantsal is now expressed at low levels in the internal part of thepiracle.Downstream of these putative Abd-B targets other genes

re activated. These include the transcription factors grainyead (grh), trachealess (trh), and engrailed (en) (Bray and

TABLE 1Expression of Posterior Spiracle Markers in Various Mutants

Markers Abd-B ct ems k

Abd-Ba NS 1 1cta,r 2 1/2 1emsr 2 1 1/2klua 2 1 1 Nnuba 2 1 1sala 2 11 1trhp 2 2 2grhr 2 2 2ena 2 1 1837p 2 11 1927p 2 2 21120p 2 2 11028p 2 2 2

Note. 1, normal expression; 2, no expression; 1/2, expression rgene; p, P element enhancer trap.

FIG. 2. Hierarchical interactions between genes expressed in thdiagram representing the inferred genetic interactions. The expresstargets depends on Abd-B but on no other gene tested. The activatio
argets. This scheme does not imply direct regulation which has only ben the secondary targets have not been tested.

afatos, 1991; Isaac and Andrew, 1996; Patel et al., 1989)nd other genes defined by enhancer trap insertions (Bellent al., 1989; Hartenstein and Jan, 1992). At this level geneegulation depends on more complex inputs. For example,hile the P1120 enhancer trap only depends on ct expres-

ion, grh and trh expression require both ems and ctfunction.

Mutants

nub sal trh grh en

1 1 1 1 11 1 1 1 11 1 1 1 11 1 1 1 1

NS 1 1 1 11 NS 1 1 11 1 1 1 NS1 1 1 1 NS1 2 NS NS NS1 2 NS NS 11 1 1 1 NS1 1 1 1 NS1 1 1 1 NS

d; 11, ectopic expression; NS, not studied; a, antibody, r, reporter

terior spiracles. The data from Table 1 have been organized in af all markers tested depends on Abd-B. The activation of the earlysecondary targets seems to be controlled by a combination of early

lu

111

S111111111

e posion on of

en proven for the ABD-B/ems interaction. Possible inputs of Abd-B


tepsfatchc

ldr

f1atrc

seitrm

mbti

apmdfsHpdst


We have analyzed the spiracle phenotypes in mutants forthe early Abd-B downstream genes (Fig. 3). In sal mutantshe stigmatophore does not form (Fig. 3B), resulting inmbryos with a normal spiracular chamber that does notrotrude. Conversely, mutations in ems and ct affect thepiracular chamber but not the stigmatophore. Mutationsor ems result in a spiracular chamber that lacks a filzkorpernd is not connected to the trachea (Fig. 3D). In ct mutantshe filzkorper is almost missing, but the trachea is stillonnected to the spiracular chamber and the spiracular

FIG. 3. Phenotype of mutants affecting the posterior spiracles.Phase-contrast images of the posterior spiracles of wild type (A) andsal (B), ct (C), ems (D), H99 (E), and trh (F) mutants. In (A) the whiterrowhead points to the stigmatophore and the black arrowheadoints to the spiracular chamber where the filzkorper forms. In salutants (B) the filzkorper forms but the stigmatophore does not

evelop. In ct mutants (C) only fragments of the filzkorper areormed and the spiracular hairs are absent. In ems mutants (D) thepiracular hairs form normally but the filzkorper does not form. In99 (E) embryos, where there is no apoptotic cell death, a normalosterior spiracle develops. In trh embryos (F), in which the tracheaoes not form, the stigmatophore is normal and the cells of thepiracular chamber invaginate, forming a filzkorper that is shorterhan the normal one.

airs are also missing (Fig. 3C). The abnormal spiracularhamber in ct mutants is not due to the absence of spiracu-


ar hairs, as the spiracular chamber is normal in embryoseleted for the achaete–scute complex that also lack spi-acular hairs (data not shown).

In trachealess mutants, where the tracheal pits do notorm and there is no tracheal network (Isaac and Andrew,996), the spiracular chamber cells still invaginate, formingfilzkorper (Fig. 3F). However, this filzkorper is shorter

han that of the wild type probably due to a secondaryequirement of trh, which is also expressed in the spiracularhamber cells.These results show that the spiracular chamber, the

tigmatophore, and the trachea develop independently ofach other. We have not been able to detect any phenotypesn mutants for either klu or nub, indicating that althoughhese genes are expressed in the spiracle, they are eitheredundant or their function is not required for spiracleorphogenesis.

Fate Map of the Spiracular Chamber at Stage 11

To understand how different mutants affect the behav-ior of the spiracle cells, we first have to analyze thewild-type development. The morphogenetic movementsthat give rise to the spiracular chamber start at st 11 afterthe tracheal pits have invaginated. The ct gene is the best

arker for the cells that will make the spiracular cham-er as it is expressed in these cells when they are still onhe surface and continues being expressed after spiraclenvagination. At st 11 ct is expressed in a group of about

70 cells arranged as a two-dimensional sheet. Most ofthese cells are posterior to the A8 tracheal pit although afew coexpress both tracheal and spiracular markers (Fig.1F). These cells are located in the dorsal half of theanterior compartment of A8 (Fig. 1G) and at this stagehave a shape similar to that of cells at homologouspositions in more anterior segments.

To follow the movements of the spiracular chamber cellsas they invaginate we have analyzed constructs made withparticular enhancers of the ct, ems, and grh genes that driveexpression of b-gal in subsets of the cells that express the ctgene at st 11 (Fig. 4). These enhancers are not driving thewhole spiracle expression of their genes but are good toolsfor studying cell specification and the morphogeneticmovements of the posterior spiracle cells. The expression ofct in the posterior spiracle is controlled by at least threedifferent enhancers (Jack and DeLotto, 1995), two of whichhave been used in this study. The ct-A4.2 enhancer (Fig. 6B)marks from st 13 the precursors of the four spiracular hairs(from now on we will refer to this enhancer as ct-four). Thect-D2.3 enhancer is expressed earlier than ct-four andmarks three groups of cells from late stage 11 (Fig. 4A). Thegrh-D4 enhancer of the grainy head (grh) gene is expressedin a single group of cells in this area (S. Bray, unpublished;Fig. 4B). The expression of ems in the spiracle is driven atleast by one enhancer: ems-1.2 (Jones and McGinnis, 1993).
This enhancer marks from st 11 a group of cells abutting thetracheal pit (Fig. 4E).

1up

peecto tructc


Double stainings of the ct-D2.3, ems-1.2, and grh-D4lacZ constructs show that they are expressed in nonover-lapping subsets of cells (Fig. 4, compare C with A and Band H with A and E). As the expression of these con-structs is maintained during the morphogenesis of thespiracular chamber, we can follow the cell movementsand observe what structures they give rise to at laterstages. At st 15, the ems-1.2-expressing cells are locatedin the deeper areas of the spiracular chamber adjacent tothe trachea. The grh-D4 cells are located over the ems-1.2-expressing cells and the ct-D2.3 are located over thegrh-D4 expressing cells, closer to the stigma opening(Figs. 4D, 4G, and 4J). The correlation of the expression ofthese three constructs at st 11 and st 15 allows us to “fatemap” the spiracular chamber primordium when it is atwo dimensional sheet of cells (Figs. 4F and 4I). Thedifferent spatial expression of these enhancers at st 11shows that the two-dimensional sheet of cells is already

FIG. 4. Cell shapes and fate map of the spiracular chamber. Exprinvagination of the spiracular chamber cells. The right column (D,G the elongated shape of the cells can be seen due to the cytoplinvaginate deeper than those marked in D and become more elongaembryo. In J the elongated shape of the cells is not detectable ahotographs (A–C, E, H) show the two-dimensional sheet of cells ombryos have been double stained with anti-en (brown) to help to lolongating. The embryo in C carries both the ct-D2.3 and grh-D4t-D2.3 and ems-1.2 constructs (compare with A and E). (F, I) schehe spiracular chamber before (F) and after (I) invagination. Yellowrange (in F), engrailed. The colored squares by the name of the consontructs in the schemes.

patterned and that the cells invaginate to precise posi-tions during development.


Formation of the Spiracular Chamber

The morphogenesis of the spiracular chamber precedesthat of the stigmatophore. As shown above with the lacZconstructs the cells can be traced as they invaginate. Theems-1.2 cells (yellow, Fig. 4F) are the first ones to invaginateand do so after the tracheal pit cells, which are located moreanteriorly, have invaginated. After them, grh-D4 cells(green, Fig. 4F) invaginate, followed later by ct-D2.3 cells(red, Fig. 4F). Thus, the invagination is not simultaneous,but proceeds sequentially from anterior to posterior. Theordered invagination results in the anterior cells of A8 beinglocated in more internal positions at stage 15 (Fig. 4I).

The invagination of the spiracular chamber is accompa-nied by drastic cell shape changes. The cell shapes can bedetected either by inducing Tau–b-gal clones (Figs. 1B andC) (A. Martı-Subirana and R. Holmgren unpublished) or bysing the ct-D2.3 or grh-D4 lacZ lines that express cyto-lasmic b-galactosidase in specific cells (Figs. 4D and 4G).

n of ems-1.2, ct-D2.3, and grh-D4 lacZ constructs before and aftershows sagittal views of invaginated spiracular chambers. In D andc location of the b-gal protein. Note that the cells marked in Gs both cells maintain a connection with the external surface of thethis construct b-gal has a nuclear location signal. All the otherspiracular chamber seen from above before it invaginates. These

the cells in the segment. At this stage the cells have not yet startedtructs (compare with A and B). The embryo in H carries both the

representation of where the different constructs are expressed inesents the expression of ems-1.2; green, grh-D4; red, ct-D2.3; ands are a key to the colors used to represent the cells expressing these

essioG, J)asmited as inf thecateconsmatic

repr

Both methods show that the cells of the spiracular chamberhave an elongated “bottle” shape, with the nucleus and


hfopdgt

b

hps


most of the cytoplasm in a basal position and a long thinapical neck extending to the external surface of the embryo.The process of elongation begins when the cell startsinvaginating and, in general, the “neck” of the bottle cell ismore elongated the further the cell invaginates.

The elongated shape of the cells within the spiracularchamber suggests that their microtubule cytoskeletonmight be polarized in the apical–basal axis. To test if this isthe case, we have compared the localization of two forms ofb-gal protein driven in the spiracular chamber with theklu-GAL4 line (Klein and Campos-Ortega, 1997). One form

as a cytoplasmic localization, while the other form is ausion with the Nod protein and localizes to the minus endsf the microtubules (Clark et al., 1997). Although bothroteins are expressed in the same cells, they have a veryifferent distribution. The cytoplasmic form labels homo-eneously the cells of the spiracular chamber; in contrast,he Nod–b-gal fusion protein localizes primarily to the

lumen of the spiracle, showing that the microtubules arepolarized with their minus ends oriented to the apical sideof the cell (Fig. 5).

The changes in cell shape could be a passive result of theforces that the cells suffer while they invaginate, or theycould be an active process by which the cells contribute tothe morphogenetic movements. To distinguish betweenthese two possibilities we have studied the shapes of thesecells in embryos in which invagination of the spiracularchamber does not occur. To achieve this we have expressedUAS-dpp driven by Arm-GAL4. In Arm-GAL4 UAS-dppembryos, although the spiracular chamber stays on thesurface, all the structures of the spiracular chamber differ-entiate (Fig. 6D). In these abnormal embryos the trachea

FIG. 5. Polarization of the spiracle chamber cells. Stage 14 em-ryos in which the klu-GAL4 line has been used to express two

different forms of b-gal protein in the posterior spiracles (arrow-eads). The first is localized in the cytoplasm and marks theosition of the bodies of the cells expressing klu-GAL4 (A). Theecond form is a fusion between Nod and b-gal and localizes to the

apical side of the cells (B), showing that the microtubules arepolarized in these cells. Note that the shape of individual cellscannot be observed due to the fact that klu-GAL4 drives expressionin most of the cells of the posterior spiracle.

invaginates and connects with the surface of the embryo.Posterior to the A8 tracheal opening, the filzkorper appears


as a flat lawn; and posterior to it the four spiracular hairs arealigned in a dorsoventral row (Fig. 6, compare A and B withD and E). Our cell shape markers show that normal cellelongation occurs in these embryos even when the spiracledoes not invaginate (Fig. 6F). This shows that cell shapechanges are not the passive result of invagination, but anactive cell process at work during spiracle morphogenesis.

Formation of the Stigmatophore

The stigmatophore development is delayed with respectto that of the spiracular chamber. The first gene activated inthe stigmatophore is the sal gene (Kuhnlein et al., 1994). Salis initially expressed at st 11 in a subset of cells in the dorsalregion of A8. This soon spreads to the posterior compart-ment and encircles the spiracular chamber cells (Fig. 7A).The expression of sal stabilizes at st 12 with no major gainor loss of sal-expressing cells in the posterior spiracle.Although the number of sal-expressing cells does notchange, the circle of sal expression changes shape drasti-cally (Figs. 7A–7D). These changes are not likely due to celldivision as, with the exception of the nervous system cells,the last mitoses in the embryo have already occurred

FIG. 6. Development of the spiracular chamber in the absence ofinvagination. Comparison of the posterior spiracle structures de-veloping in two versus three dimensions. Cuticles or embryosstained with markers in wild type (A–C) or noninvaginated spi-racles (D–F). The cuticular structures present in the wild-typespiracular chamber are present in spiracles that have not invagi-nated, showing that the specification of different cell types canoccur in the absence of normal morphogenetic movements (com-pare A and D). The trachea in these embryos invaginates normally,and the cells of the spiracular chamber stay on the surface, wherethey generate a flat filzkorper (arrow). Other elements such as thetwo vesicle-like structures normally present at the stigma form onone end of the filzkorper (arrowheads), and the spiracular hairs formin a line posterior to the uninvaginated filzkorper. A ct-four linestaining the precursors of the spiracular hairs shows the four cellsaround the stigma (B); these cells are aligned in the uninvaginatedspiracle (E). A grh-D4 line showing the cell shape after invagination
(C) and an embryo in which although the invagination does notoccur the cells still elongate (F).

a7b

in b


(Campos-Ortega and Hartenstein, 1997). To analyze whythe circle of sal changes shape, we have recorded thenumber and position of sal-expressing nuclei in the stig-

FIG. 7. Cell rearrangement of the sal-expressing cells in the stigmst 14 (C), and st 15 (D) (magnification in D is half of that in A–C). Tand C9 delimit the sal-expressing cells in the stigmatophore. Thinvaginates to form the spiracular chamber. Outside the stigmatopstigmatophore, sal is expressed initially in a two-cell-wide incomplthe cells of the spiracular chamber invaginate, the semicircle close3-cell-wide circle at st 13 (B) becomes 4 to 5 cells wide at stage 14 (Cthe width of the sal circle). Note that due to the rearrangementcircumference simultaneously decreases (The circumference is reprrearrangement in the stigmatophore. These data have been obtaineB9 and C9. The green line represents the number of nuclei in the widchamber, and the blue line the circle of cells most distal to the spiare displaced to central positions, decreasing the number of nucleierror.

matophore at different stages of development (Fig. 7E). By st12 sal is expressed in a circle two to three cells wide, with

c7


n internal circumference of approximately 25 cells (Fig.A). At later stages the width of the sal circle of expressionroadens, while simultaneously the inner and outer cir-

hore. Expression of sal in the stigmatophore at st 12 (A), st 13 (B),osition of A8 is indicated by a braket. The blue and red lines in B9a inside the incomplete circle of sal expression in A8 (asterisk), sal is also expressed in the trachea (arrowhead) and in A9. In theircle around the cells that will form the spiracular chamber (A). Asand the cells start rearranging their positions. As a result, the 2- tod about 10 cells wide at stage 15 (green line in B9 and C9 representsell position, as the width increases, the number of cells on theed in B9 and C9 by the red and blue lines.) (E) Graph illustrating cellcounting the number of nuclei along the three lines represented inthe sal circle, the red line the circle of cells closer to the spiracular

ar chamber. Cells that initially where along the red and blue linesoth circles and increasing the width. Bars in E represent standard

atophe p

e arehoreete cs (B)

) anin c

esentd byth ofracul

umferences become smaller (compare Fig. 7C with Fig.B). The decrease of nuclei in the inner and outer circum-


nmrp

sriati8tcngantszccss

n thee ct-


ference cannot be accounted for by cell death, as in homozy-gous embryos for the H99 deletion in which apoptosis does

ot occur (White et al., 1994) the stigmatophore develop-ent is normal (Fig. 3E). These results show that cell

earrangements are responsible for closing the stigmato-hore and, indirectly, elongating the stigmatophore.

Formation of a Continuous Tracheal–SpiracularLumen

For the respiratory system to function, the lumen of thespiracular chamber and the trachea must connect. Thebranch that connects the dorsal trunk of the trachea withthe posterior spiracle is the spiracular branch of A8 (Man-ning and Krasnow, 1993). In segments where the spiraclesdo not form, the spiracular branch stays attached to theepidermis of the embryo and the lumen collapses.

To study how the spiracular branch of A8 links to theposterior spiracle we have used an antibody against theCrumbs protein (anti-crb) and the markers described above.The Crumbs protein localizes to the apical surface of allepidermal cells and can be used as a marker of the lumen ofthe trachea and the spiracle (Tepass et al., 1990).

When the trachea primordia first invaginate there are nosigns of spiracle morphogenesis. At this point, the stigma(opening of the respiratory system) is formed by the hole in

FIG. 8. The invagination of the spiracular chamber proceeds froinvagination in the sagittal view (A–D) or the dorsal view (E–J). Amarks the lumen of the tracheal system, and anti-b-gal to detect difst 14 (B). (C–D) The expression of ct-D2.3 at stages similar to those Awhere the stigma opening is located at st 11. Note that at st 14 theclose to the stigma (A) but at later stages are deep inside and away faway from the stigma (C) but at st 14 are closer to it (D). The tracheshown by the crumbs staining is due to the trachea going out of tharrowhead points at the stigma. In these panels both crumbs andinvaginate the stigma slides back toward its final position. Whedisplaced ventrally, forming a semicircle (H and I). Afterward (J) th

the center of the invaginated tracheal pit of A8 and nospiracular lumen exists. The spiracle cells are still on the


urface of the embryo and only later start expressing spi-acle specific genes that induce the invagination. Thenvagination of the spiracular chamber cells occurs fromnterior to posterior. The spiracular chamber cells closer tohe trachea (those expressing ems-1.2, yellow in Fig. 4F)nvaginate attached to the neighbouring tracheal cells (Figs.A and 8B). At the same time, the stigma shifts back fromhe tracheal pit to the ems-1.2-expressing cells. This pro-ess spreads posteriorly and as more posterior cells invagi-ate, the stigma keeps shifting posteriorly first to therh-D4, and then ctD-2.3 expressing cells (Figs. 8C and 8Dnd 8E–8J) until it reaches its final position. During invagi-ation, the cells of the trachea and the spiracle never looseheir connection. The formation of the common tracheal–piracular lumen can be compared to the movement of aipper: the stigma shifts back its position between differentells and the invagination process brings into proximityells that initially were distant in the two-dimensionalheet. The posterior spiracle lumen is formed by the apicalurfaces of the cells that invaginate.

Genes Required for the Connection of the PosteriorSpiracle to the Trachea

We have studied if the genes required for the tracheal andspiracular lumens to connect are different than the genes

terior to posterior. View of the spiracular chamber region duringbryos have been doubly stained with anti-crumbs (black), which

t spiracle markers. (A, B) The expression of ems-1.2 at st 11 (A) andB respectively. The arrowheads in A–B and C–D point to the placea has moved posteriorly. The ems-1.2 expressing cells are initiallyit (B). This contrasts with the ct-D2.3 cells that at st 11 are farthermen is in all cases connected to the stigma, the apparent dead endal plane. In E–J the focus is on the surface of the embryo and the2.3 are stained black. Note that as the spiracular chamber cellsanterior cells invaginate the dorsal ct-D2.3-expressing cells get

D2.3 cells invaginate themselves (see Fig. 4D for a later stage).

m anll emferen

andstigmromal lue focct-D

required to form the continuous tracheal tree. In mutantsfor the Drosophila FGF and FGF-R homologues branchless


ct

pattts

ostftcwetmc

batc


and breathless (btl) the tracheal pits invaginate but as theydo not migrate toward each other, they do not form acontinuous network (Klambt et al., 1992; Sutherland et al.,1996). In contrast, in btl mutants the posterior spiracleonnects normally to the A8 spiracular branch of therachea.

In mutants for Abd-B the stigma of A8 does not slideosteriorly, but stays in the same position as in anteriorbdominal segments, where the spiracular branch attacheso the outside epidermis (Fig. 9B). Because of this, we haveested if the downstream targets of ABD-B are required forhe connection of the trachea to the posterior spiracles. Thepiracle–trachea connection occurs in ct and sal mutants

but not in ems mutants. To detect the behavior of the cellsof the spiracular chamber and the movements of the stigmain ems mutant embryos, we have used the lacZ markers andthe anti-crb antibody. In ems mutants the cells expressingct-D2.3 and grh-D4 invaginate at the expected moment,acquiring the normal elongated cell shape (not shown).

FIG. 9. Formation of the spiracle–trachea lumen in differentmutants. Lateral views of stage 13 embryos stained with anti-crb toshow the lumen of the spiracle and the trachea. (A) Wild-typeembryo with the posterior spiracle forming in the dorsal side of A8(black arrowhead). At this stage the lumen is well formed in thecenter of the spiracle (white arrowhead). Note that the expressionof crb is stronger in the spiracle lumen than in the tracheal lumen.In Abd-B mutant embryo (B), the posterior spiracle has not formed(the black arrowhead points to the dorsal position where it shouldbe located) there is no spiracle lumen and the trachea opens to thesurface directly through the spiracular branch (white arrowhead).Embryos mutant for ems (C–D) stained for crb (C) or carrying theems-1.2 reporter construct (D). The expression of the construct ismuch weaker than that in the wild type. Two groups of cellsexpressing ems-1.2 have not invaginated and are on the surface(white arrowhead points to the anterior group). The posteriorspiracles (black arrowhead) have not formed a lumen and thetracheal opening is outside of the spiracle (C, white arrowhead).

However, anti-crb antibody staining shows that in emsmutants the stigma is stalled where the spiracular branch


riginated and does not slide posteriorly, a phenotypeimilar to that of Abd-B mutants (Fig. 9C). This results inhe stigma being outside of the spiracle. To determine theate of the cells that at st 11 were immediately adjacent tohe tracheal pit in ems mutants we have used the ems-1.2onstruct. In ems mutants the ems-1.2 marker stainseaker than in wild type embryos. However, the cells that

xpress lacZ do not invaginate and remain on the surface ofhe embryo (Fig. 9D), suggesting that what stalls the move-ent of the stigma is the lack of invagination of the spiracle

ells that are in contact with the trachea.As in ems mutants the stigma does not move posteriorly,

ut the ct-D2.3- and grh-D4-expressing cells can invaginatend change cell shape, these results show that cell elonga-ion and the formation of a lumen are two independentlyontrolled processes.

DISCUSSION

The posterior spiracles present several advantages for thestudy of morphogenesis: they are conspicuous and easy toobserve; they develop fast; and there are many molecularmarkers that can be used to identify the cells of the spiraclebefore they are different from their neighbors, allowingtheir movements to be followed during the morphogeneticprocess. As the posterior spiracles are formed at 15 h ofdevelopment, we can study most mutations in genes in-volved in the control of morphogenesis as they allow theembryo to develop beyond this stage.

Morphogenetic Mechanisms Required for theFormation of the Posterior Spiracles

Cell proliferation and cell death are two major morpho-genetic mechanisms (Conlon and Raff, 1999; Vaux andKorsmeyer, 1999). Neither appears to play a major role inposterior spiracle formation. This is shown by the fact thatmutations that abolish apoptosis make a normal spiracleand that the morphogenesis of the posterior spiracle startsat late st 11, after the last mitoses in the embryonicepidermis have finished (Campos-Ortega and Hartenstein,1997). On the other hand, control of cell shape and cellrearrangements play an important role in the formation ofthe spiracular chamber and the stigmatophore.

We have shown that cell elongation occurs simulta-neously to the invagination of the spiracular chamber cells.The cell shape changes are not due to mechanical forcesimposed on the invaginating cells, as in embryos in whichinvagination does not occur, the cells change shape at theappropriate time of development. The elongation does notoccur simultaneously in all cells, but starts in the moreanterior ones (Figs. 10A–10D) and, in general, the invagi-nating cells keep contact with the external surface of theembryo. This results in the cells that have invaginated
earlier being deeper in the spiracular chamber and longer(compare Figs. 4D and 4G). The elongated cells have polar-

tTtmatt

siccwwhmt

tsiv1dsdaa

later


ized microtubules with their minus ends toward the apicalsurface. We cannot at this point say if the polarization ofthe microtubules is the cause of the elongation of thesecells or if it is required for the stabilization of the shape.Other cytoskeletal components are probably also importantto achieve the cell shape, as stainings with phaloidin–rhodamine show that high levels of actin accumulate in thenarrow processes of the bottle cells (data not shown).

The invagination of the cells starting in the anterior partof the spiracular chamber primordium and spreading poste-riorly suggests the involvement of a relay signaling mecha-nism. However, several experiments indicate that the well-timed invagination is controlled cell autonomously bytranscription factors activated in the two-dimensional pri-mordium. First, in mutants for the transcription factor emshe cells neighboring the tracheal pit do not invaginate.his stops the progression of the lumen but does not affect

he invagination of more posterior cells. Second, geneticarkers specific for different cells of the spiracular chamber

re activated in the two-dimensional primordium before

FIG. 10. Model for the morphogenetic movements of the spiraculaof the posterior spiracle at different stages of development. The upview from the top (anterior, left; dorsal, up). The cells are color codgreen; and red, cells located at different positions in the spiracularcells have invaginated and still connect to the outside ectoderm; th(B and F) the cells of the trachea are not on the surface any more aare invaginating. The invagination of the trachea inside the embryosurrounding cells filling the space previously occupied by the invathe stigmatophore (dark green) and the ventral displacement of thchamber cells elongate and invaginate (H and I). The cells that icontinue elongating becoming longer than those that invaginateelevation over the surrounding ectoderm (D and I).

he onset of the morphogenetic movements. Therefore, inhe posterior spiracle, signaling might be important for the

K


patial activation of the expression of transcription factorsn the two-dimensional primordium, and these result in theontrol of cell elongation autonomously. However, weannot discard the alternative possibility that a signalingave spreads through the primordium, informing the cellshen to activate the invagination mechanisms. In thisypothesis ems mutant cells would have the invaginationechanism blocked, but they would be capable of relaying

he signal to more posterior cells.The formation of elongated bottle cells is common during

he invagination of tissues in several organisms. In Dro-ophila, bottle-shaped cells form transiently during thenvagination of the tracheal pits and in the formation of theentral furrow during mesoderm invagination (Costa et al.,994). The major difference of these cells and those weescribe here is that, in the posterior spiracle the bottle cellhape becomes stabilized. Bottle-shaped cells have beenescribed in the dorsal lip of the blastopore in amphibiansnd in the vegetal plate of the sea urchin where these cellsre critical for the initiation of invagination (Keller, 1981;

mber and the stigmatophore. Schematic representation of the cellsow represents a sagittal section (anterior left) and the lower row arange, trachea cells; dark green, stigmatophore cells; yellow, lightber (compare to Figs. 4F and 4I). At stage 11 (A and E), the tracheairacle cells are all of similar shape and are on the surface. At st 12e cells most proximal to the trachea (yellow) have elongated andthe smaller apical surface of the cells distorts the tissue, with theed cells (F and G). This is noticeable in F and G by the closure ofrsal ct-D2.3 cells (red). As development proceeds more spiracularnated earlier are still attached to the surface of the embryo and(C). The cell rearrangement in the stigmatophore results in its

r chaper red: o

chame spnd thand

ginate do

nvagi

imberly and Hardin, 1998).The morphogenetic mechanism used for the formation of


1scmfeirtsc

1inArm


the stigmatophore is completely different. The cells do notchange shape very markedly but rearrange their positions inthe epithelium in a mechanism which is similar to theconvergent extension movements described during the gas-trulation of organisms ranging from vertebrates such asXenopus or zebra fish, to Caenorhabditis (Warga and Kim-mel, 1990; Williams-Masson et al., 1998; Wilson and Keller,991). During convergent extension, cells that are side byide intercalate, changing their positions so that one be-omes anterior to the other (Figs. 10F–10I). This rearrange-ent results in the elongation of the epithelium that will

orm the stigmatophore. At the same time, convergentxtension in the stigmatophore solves the problem that thenvagination of the spiracular chamber causes to the integ-ity of the epithelium: cell rearrangement is the solutionhe stigmatophore cells use to make a smaller contacturface with the shrinking apical surfaces of the spiracularhamber cells.

The Patchwork Formation of the Posterior SpiracleCould Be a Result of a Step by Step Evolution

Many insects have one pair of spiracles per segment fromT2 to A8, but this number has varied during evolution andalso during development (Keilin, 1944). For example, whilein the larva of Drosophila there are only two pairs, posteriorspiracles in A8 and anterior spiracles in T1, most segmentsof the adult have functional spiracles. In the larvae of otherinsect species, the number of functional spiracles varies (0,1, 2, 8, 9, or 10), and in many cases this variation isconsidered to be an adaptation to the physiological require-ments of the larva to the particular environment in which itdevelops (Keilin, 1944). In the genus Drosophila the shapeand size of spiracles can vary. Some species having similarshapes as Drosophila melanogaster and others being longeror shorter (Wheeler, 1987). This variability suggests that theshape of the posterior spiracles is under strong selectivepressure and that the morphology they now have is arelatively recent adaptation.

The trachea and the posterior spiracle develop indepen-dently, each using a different morphogenetic mechanismthat is regulated by independent genetic pathways. Forexample, trachea formation does not require Abd-B but isdependent on the trachealess gene for the invagination;while for the formation of the posterior spiracle, trachealessplays a secondary role and Abd-B is fundamental. The sameindependent development applies to the external and inter-nal parts of the posterior spiracle. The spiracular chamberrequires ems and ct but it does not express sal while theopposite is true in the stigmatophore. All three genes areexpressed in many other tissues (Blochinger et al., 1990;Dalton et al., 1989; Kuhnlein et al., 1994; Walldorf andGehring, 1992) and have been coopted in the spiracledownstream of Abd-B. This patchwork formation for astructure that has to be perfectly connected to be functional
relies on a perfect integration of the three different geneticpathways and morphogenetic mechanisms. This integra-

tion might have been achieved by a step by step evolution.Initially the spiracular branch opening could act as a stigmafor air exchange. Later on in evolution, using a differentgenetic pathway, the adjacent cells could be recruited toform filter-like structures as the filzkorper. Finally, a lastgroup of cells could be recruited to form the periscopicstructure of the stigmatophore. This scenario would explainthe patchwork development of the posterior spiracle wenow see, as at each of these stages a different morphoge-netic mechanism would have been used with its controlbeing under the regulation of a new subset of transcriptionfactors.

The Posterior Spiracles Gene Regulatory Cascade

Abd-B is necessary and sufficient to induce posteriorspiracle development in the dorsal part of the trunk of thelarva. In Abd-B mutants the posterior spiracles do not formand, conversely, ectopic expression of ABD-B protein inanterior segments results in ectopic posterior spiracles(Castelli-Gair et al., 1994; Kuziora, 1993; Lamka et al.,992; Sanchez-Herrero et al., 1985). These experimentsndicate that Abd-B regulates all the morphogenetic mecha-isms that lead to the formation of the posterior spiracle.bd-B is expressed on A8 both in the ventral and dorsal

egion, so there must be some positional information thatodifies the transcriptional outcome of Abd-B expression.

We can only guess that the interaction of the ABD-B proteinwith cofactors expressed differentially in a dorsoventralpattern or an interaction with signaling molecules requiredfor dorsoventral patterning such as the TGF-b homologdecapentaplegic (dpp) will give the intrasegmental differ-ences.

The first genes that we have found to be expressed in aposterior spiracle restricted pattern are a group of earlyresponsive transcription factors whose expression does notdepend on each other but requires ABD-B. We call thesegroup early targets and, though only ems has been shown tobe a direct target (Jones and McGinnis, 1993), their timingof expression suggests that they could be regulated directlyby ABD-B at the transcriptional level. The expression of theearly targets ct, ems, klu, nub is restricted to the primor-dium of the spiracular chamber and with a slight delay thesal gene becomes expressed in the surrounding cells thatwill become the stigmatophore. This is the first indicationof the existence of two distinct populations of cells in thespiracle suggesting that it is at this stage that the spiracle ispatterned into stigmatophore and spiracular chamber. Sec-ondary targets seem to be either expressed in the stigmato-phore or in the spiracular chamber, in agreement with theobservation that both cell populations develop using differ-ent morphogenetic strategies.

With the available data, the genetic network of thestigmatophore seems to be a simple cascade downstream ofsal. The genetic network of the spiracular chamber shows a
more complicated relationship with several early transcrip-tion factors acting in parallel. This richness of transcription

B

B

B


factors could provide the positional information requiredfor the precise spatial activation of the cell markers ex-pressed when the spiracular chamber primordium is stilltwo dimensional (Fig. 4F).

In summary, in this paper we have described the devel-opment of an essential part of the tracheal system: theposterior spiracles. The spiracles are a good model to studythe cellular and molecular mechanisms controlling cellshape and cell rearrangements, two mechanisms which areused during the morphogenesis of a variety of organisms.The relative simplicity and experimental accessibility ofthe posterior spiracles of Drosophila should help to under-stand how these basic processes are controlled.

ACKNOWLEDGMENTS

We thank S. Celniker, Y. N. Jan, R. Schuh, P. Lawrence, T. Klein,W. McGinnis, D. Andrew, S. Bray, J. Jack, J. F. de Celis, H. Steller,E. Knust, and the Tubingen and Bloomington stock centers forantibodies and flies; M. Akam, M. Bate, S. Brown, and E. MartınBlanco for helpful discussions and comments to the manuscript; A.Martı-Subirana and R. Holmgren for the Tau lacZ flip-out cassette;and M. Gonzalez-Gaitan for inducing the Tau lacZ clones. Weespecially thank M. Akam and A. Martınez Arias for their supportat the early stages of this work. This work was supported by TheRoyal Society and The Wellcome Trust.

REFERENCES

Bellen, H. J., O’Kane, C. J., Wilson, C., Grossniklaus, U., Pearson,R. K., and Gehring, W. J. (1989). P-element-mediated enhancerdetection: A versatile method to study development in Drosoph-ila. Genes Dev. 3, 1288–1300.

ier, E., Vaessin, H., Shepherd, S., Lee, K., McCall, K., Barbel, S.,Ackerman, L., Carretto, R., Uemura, T., Grell, E., Jan, L., and Jan,Y. (1989). Searching for pattern and mutation in the Drosophilagenome with a P-lacZ vector. Genes Dev. 3, 1273–1287.

lochinger, K., Bodmer, R., Jan, L. Y., and Jan, Y. N. (1990). Patternsof expression of Cut, a protein required for external sensory organdevelopment in wild-type and cut mutant Drosophila embryos.Genes Dev. 4, 1322–1331.

lochlinger, K., Bodmer, R., Jack, J., Jan, L. Y., and Jan, Y. N. (1988).Primary structure and expression of a product from cut, a locusinvolved in specifying sensory organ identity in Drosophila.Nature 333, 629–635.

Botas, J. (1993). Control of morphogenesis and differentiation byHOM/Hox genes. Current Opinions Cell Biol. 5, 1015–1022.

Brand, A. H., and Perrimon, N. (1993). Targeted gene expression asa means of altering cell fates and generating dominant pheno-types. Development 118, 401–415.

Bray, S. J., and Kafatos, F. C. (1991). Developmental function ofElf-1: An essential transcription factor during embryogenesis inDrosophila. Genes Dev. 5, 1672–1683.

Campos-Ortega, J. A., and Hartenstein, V. (1997). “The EmbryonicDevelopment of Drosophila melanogaster,” Springer-Verlag,Berlin.

Casanova, J., Sanchez-Herrero, E., and Morata, G. (1986). Identifi-cation and characterisation of a parasegment specific regulatory
element of the Abdominal-B gene of Drosophila. Cell 47, 627–636.

Castelli-Gair, J., Greig, S., Micklem, G., and Akam, M. (1994).Dissecting the temporal requirements for homeotic gene func-tion. Development 120, 1983–1995.

Celniker, S. E., Keelan, D. J., and Lewis, E. B. (1989). The moleculargenetics of the bithorax complex of Drosophila: Characterizationof the products of the Abdominal-B domain. Genes Dev. 3,1424–1436.

Chen, P., Nordstrom, W., Gish, B., and Abrams, J. M. (1996). grim,a novel cell death gene in Drosophila. Genes Dev. 10, 1773–1782.

Clark, I. E., Jan, L. Y., and Jan, Y. N. (1997). Reciprocal localizationof Nod and kinesin fusion proteins indicates microtubule polar-ity in the Drosophila oocyte, epithelium, neuron and muscle.Development 124, 461–470.

Conlon, I., and Raff, M. (1999). Size control in animal development.Cell 96, 235–244.

Costa, M., Wilson, E. T., and Wieschaus, E. (1994). A putative cellsignal encoded by the folded gastrulation gene coordinates cellshape changes during Drosophila gastrulation. Cell 76, 1075–1089.

Dalton, D., Chadwick, R., and McGinnis, W. (1989). Expression andembryonic function of empty spiracles: A Drosophila homeo boxgene with two patterning functions on the anterior-posterior axisof the embryo. Genes Dev. 3, 1940–56.

de Celis, J. F., Barrio, R., and Kafatos, F. C. (1996). A gene complexacting downstream of dpp in Drosophila wing morphogenesis.Nature 381, 421–424.

Graba, Y., Aragnol, D., and Pradel, J. (1997). Drosophila Hoxcomplex downstream targets and the fuction of homeotic genes.BioEssays 19, 379–388.

Grether, M. E., Agapite, J., White, K., and Steller, H. (1995). Thehead involution defective gene of Drosophila melanogasterfunctions in programmed cell death. Genes Dev. 9, 1694–1708.

Hartenstein, V., and Jan, Y. N. (1992). Studying Drosophila em-bryogenesis with P-lacZ enhancer trap lines. Roux’s Arch. Dev.Biol. 201, 194–220.

Isaac, D. D., and Andrew, D. J. (1996). Tubulogenesis in Drosophila:A requirement for the trachealess gene product. Genes Dev. 10,103–117.

Jack, J., and DeLotto, Y. (1995). Structure and regulation of acomplex locus: The cut gene of Drosophila. Genetics 139,1689–1700.

Jan, Y. N., and Jan, L. Y. (1993). The peripheral nervous system. In“The development of Drosophila melanogaster” (A. MartıntezArias, Ed.), Vol. 2, pp. 1207–1244. Cold Spring Harbor LaboratoryPress, Cold Spring Harbor, NY.

Jones, B., and McGinnis, W. (1993). The regulation of emptyspiracles by Abdominal-B mediates an abdominal segment iden-tity function. Genes Dev. 7, 229–240.

Keilin, D. (1944). Respiratory systems and respiratory adaptationsin larvae and pupae of Diptera. Parasitology 36, 1–66.

Keller, R. E. (1981). An experimental analysis of the role of bottlecells and the deep marginal zone in gastrulation. J. Exp. Zool.216, 81–101.

Kimberly, E. L., and Hardin, J. (1998). Bottle cells are required forthe initiation of primary invagination in the sea urchin embryo.Dev. Biol. 204, 235–250.

Klambt, C., Glazer, L., and Shilo, B.-Z. (1992). breathless, a Dro-sophila FGF receptor homolog, is essential for migration oftracheal and specific midline glial cells. Genes Dev. 6, 1668–1678.

Klein, T., and Campos-Ortega, J. A. (1997). klumpfuss, a Drosophilagene encoding a member of the EGR family of transcription


M

N

P

S

S

S

T

T

V

W

W

W

W

W

W

Y

Y


factors, is involved in bristle and leg development. Development124, 3123–3134.

Kuhnlein, R. P., Frommer, G., Friedrich, M., Gonzalez-Gaitan, M.,Weber, A., Wagner-Bernholz, J. F., Gehring, W. J., Jackle, H., andSchuh, R. (1994). spalt encodes an evolutionary conserved zincfinger protein of novel structure which provides homeotic genefunction in the head and tail region of the Drosophila embryo.EMBO J. 13, 168–179.

Kuziora, M. A. (1993). Abdominal-B protein isoforms exhibitdistinct cuticular transformations and regulatory activities whenectopically expressed in Drosophila embryos. Mech. Dev. 42,125–137.

Lamka, M. L., Boulet, A. M., and Sakonju, S. (1992). Ectopicexpression of UBX and ABD-B proteins during Drosophila em-bryogenesis: Competition, not a functional hierarchy, explainsphenotypic suppression. Development 116, 841–854.

Leiss, D., Hinz, U., Gasch, A., Mertz, R., and Renkawitz-Pohl, R.(1988). b3 tubulin expression characterises the differentiatingmesodermal germ layer during Drosophila embryogenesis. De-velopment 104, 525–531.anning, G., and Krasnow, M. A. (1993). Development of theDrosophila tracheal system. In “The Development of Drosophilamelanogaster” (M. Bate and A. Martınez Arias, Ed.), Vol. 1, pp.609–685. Cold Spring Harbor Laboratory Press, Cold SpringHarbor, NY.g, M., Dıaz-Benjumea, F., and Cohen, S. M. (1995). nubbinencodes a POU-domain protein required for proximal distalpatterning in the Drosophila wing. Development 121, 589 –599.

atel, N. H., Martın-Blanco, E., Coleman, K. G., Poole, S. J., Ellis,M. C., Kornberg, T. B., and Goodman, C. S. (1989). Expression ofengrailed proteins in arthropods, annelids, and chordates. Cell58, 955–968.

anchez-Herrero, E., Vernos, I., Marco, R., and Morata, G. (1985).Genetic organisation of the Drosophila bithorax complex. Na-ture 313, 108–113.

anson, B., White, P., and Vincent, J.-P. (1996). Uncouplingcadherin-based adhesion from wingless signalling in Drosophila.Nature 383, 627–630.

taehling-Hampton, K., and Hoffmann, F. M. (1994). Ectopic deca-pentaplegic in the Drosophila midgut alters the expression of
five homeotic genes, dpp, and wingless, causing specific morpho-logical defects. Dev. Biol. 164, 502–512.

abata, T., Schwartz, C., Gustavson, E., Ali, Z., and Kornberg, T. B.(1995). Creating a Drosophila wing de novo, the role of engrailed,and the compartment border hypothesis. Development 121,3359–3369.epass, U., Theres, C., and Knust, E. (1990). The Drosophila genecrumbs encodes an EGF-like protein expressed on apical mem-branes of Drosophila epithelial cells and required for organisa-tion of epithelia. Cell 61, 787–799.aux, D. L., and Korsmeyer, S. J. (1999). Cell death in development.Cell 96, 245–254.alldorf, U., and Gehring, W. J. (1992). Empty spiracles, a gap genecontaining a homeobox involved in Drosophila head develop-ment. EMBO J. 11, 2247–2259.arga, R. M., and Kimmel, C. B. (1990). Cell movements duringepiboly and gastrulation in zebrafish. Development 108, 569–580.heeler, M. R. (1987). Drosophilidae. In “Manual of NearticDiptera” (Biosystematics Research Centre, Ed.), Vol. 2, pp. 1011–1018. Canada Communication Group Publishing, Ottawa, On-tario.hite, K., Grether, M., Abrams, J. M., Young, L., Farrell, K., andSteller, H. (1994). Genetic control of programmed cell death inDrosophila. Science 264, 677–683.illiams-Masson, E. M., Heid, P. J., Lavin, C. A., and Hardin, J.(1998). The cellular mechanism of epithelial rearrangementduring morphogenesis of the Caenorhabditis elegans dorsalhypodermis. Dev. Biol. 204, 263–276.ilson, P., and Keller, R. (1991). Cell rearrangement during gastru-lation of Xenopus: Direct observations of cultured explants.Development 112, 289–300.ang, X., Bahri, S., Klein, T., and Chia, W. (1997). Klumpfuss, aputative Drosophila zinc finger transcription factor, acts todifferentiate between the identities of two secondary precursorcells within one neuroblast lineage. Genes Dev. 11, 1396–1408.

eo, S. L., Lloyd, A., Kozak, K., Dinh, A., Dick, T., Yang, X.,Sakonju, S., and Chia, W. (1995). On the functional overlapbetween two Drosophila POU homeo domain genes and the cellfate specification of a CNS neural precursor. Genes Dev. 9,1223–1236.

Received for publication April 13, 1999
Revised June 22, 1999
Accepted June 22, 1999


study of the posterior spiracles of drosophila as a model ... · model to understand the genetic...

Documents