Drosophila Hox and Sex-Determination Genes Control Segment Elimination through EGFR and extramacrochetae Activity

The formation or suppression of particular structures is a major change occurring in development and evolution. One example of such change is the absence of the seventh abdominal segment (A7) in Drosophila males. We show here that there is a down-regulation of EGFR activity and fewer histoblasts in the male A7 in early pupae. If this activity is elevated, cell number increases and a small segment develops in the adult. At later pupal stages, the remaining precursors of the A7 are extruded under the epithelium. This extrusion requires the up-regulation of the HLH protein Extramacrochetae and correlates with high levels of spaghetti-squash, the gene encoding the regulatory light chain of the non-muscle myosin II. The Hox gene Abdominal-B controls both the down-regulation of spitz, a ligand of the EGFR pathway, and the up-regulation of extramacrochetae, and also regulates the transcription of the sex-determining gene doublesex. The male Doublesex protein, in turn, controls extramacrochetae and spaghetti-squash expression. In females, the EGFR pathway is also down-regulated in the A7 but extramacrochetae and spaghetti-squash are not up-regulated and extrusion of precursor cells is almost absent. Our results show the complex orchestration of cellular and genetic events that lead to this important sexually dimorphic character change.


Introduction
A major change during evolution is the disappearance of a particular organ or structure. This event is sometimes restricted to one sex, and therefore needs the coordination between sexdetermination genes and those that control pattern [1], such as the Hox genes, a group of genes that specify different structures along the antero-posterior axis [2]. One example of such coordination is the control of pigmentation in the Drosophila melanogaster posterior abdomen, uniformly pigmented in males but not in females. This character depends on the Hox gene Abdominal-B (Abd-B), required in abdominal (A) segments A5-A9 [3][4][5], and whose protein levels increase gradually and posteriorwards in each segment [6,7], and also on the two protein isoforms of the sex-determining gene doublesex (dsx): DsxM (in males) and DsxF (in females). The combined activity of these proteins and Abd-B promotes the development of sex-specific pigmentation [8,9]. Another significant morphological difference between Drosophila males and females is the seventh abdominal segment (A7), absent in males and present in females. Abdominal segments derive from histoblast nests, groups of cells that intermingle with cuticular larval epidermal cells (LECs), are quiescent during the larval period and proliferate rapidly at the beginning of the pupal period [10][11][12][13]. There are four histoblast nests in each hemi-segment: two dorsal (anterior, a, which forms the dorsal part of the abdominal cuticle, the tergite, and posterior, p), one ventral, developing the ventral region (the sternite) and part of the lateral region (the pleura), and one making the spiracle [12,13]. When pupation starts, the histoblasts proliferate and spread, whereas the larval epidermal cells that are contiguous to them die and are extruded, until the whole abdominal region is covered by the histoblasts, which secrete the adult cuticle [11][12][13][14].
The study of the elimination of the male A7 has been recently addressed [15]. In this analysis, it was demonstrated that the absence of wingless (wg) expression in the male A7 segment contributes to the disappearance of this metamere, probably by regulating cell proliferation. It was also shown that the forced expression of a ligand of the Epidermal growth factor receptor (EGFR) pathway, vein, makes a small A7 in the male and that segment compartmental transformation (from A7p to A6a) and restricted apoptosis also contribute to the sexual dimorphism in this segment.
We have studied the mechanisms of male A7 elimination and report here that at early pupa there is less number of histoblasts in the male A7 due, at least in part, to the down-regulation by Abd-B of the activity of the EGFR pathway; at later pupal stages the A7 histoblasts undergo extrusion under the control of the HLH protein Extramacrochetae (Emc). Abd-B regulates dsx expression in males and females, but only DsxM drives the massive extrusion of male A7. Our results show that different cellular events, under the joint regulation of the sex-determination pathway and Hox activity, underlie the disappearance of a particular structure.

Results
The different development of the Drosophila A7 in males and females ( Figure S1A, B) depends on Abd-B [3,4] and on the sexdetermination pathway [16] (Figure S1C, D). We have studied the elimination of the male A7 tergite by comparing the behavior of dorsal histoblasts in the A6 (which remains) and the A7 (which disappears). The expression driven by a escargot (esg)-Gal4 line [17] in the abdomen marks specifically the histoblasts, which can be also distinguished from surrounding LECs because they are diploid (and small) whereas the LECs are polytenic (and big). To permanently label histoblasts we have used a genetic combination that we name (p)esg-Gal4 [12] (see Materials and Methods].

EGFR pathway activity is reduced in the male A7
At the end of the larval stages the number of histoblasts in the male A7a and A7p dorsal nests is similar to the corresponding nests of the A6 ( [11]; and data not shown). In the first 10h after puparium formation (APF) the histoblasts undergo three nearlysynchronic divisions without cell growth, thus defining the first phase of histoblast pupal development [12,13,18,19]. Time-lapse movies show that during this phase the A6a and A7a nests, which develop into the A6 and A7 tergites, respectively, show similar cell division rates, with just a small delay in the A7a nest.
In a second phase, from ,10 h to ,35 h APF, the histoblasts divide asynchronously and cell division is accompanied by cell growth [12,13,19]. Histoblasts also spread in the epidermis and replace LECs [12,14]. During this phase the male A7 histoblasts undergo fewer cell divisions than those of the A6 ( [15]; and our observations) so that their number at about 24-29 h APF is smaller than that of the A6 (Video S1; Figure 1A-A0, 1J, J9). We also note that the size of the A7 histoblasts is bigger than that of the A6 histoblasts ( Figure 1J, J9; Figure S1E-E0).
Abd-B levels are higher in the pupal A7 than in the A6 [8] (Figure S1G, G9). By transforming the A6 into the A7 with the Abd-B Fab7-1 mutation [20,21], we observed a concomitant change in Abd-B levels, cell number and cell size ( Figure S1G-H9). To study the reciprocal transformation we used a Gal4 line (MD761-Gal4) that is inserted within the infraabdominal-7 (iab-7) region of this gene (position between 3R:12,725,043 and 3R:12,725,044, Flybase), close to or within the Fab-7 boundary [20][21][22]. The iab-7 regulatory domain activates Abd-B in parasegment 12 (A6p-A7a) [20,21]. In accordance with its location, MD761-Gal4 drives expression of UAS constructs in this parasegment and some posterior cells (M. Calleja and G. Morata, personal communication; Figure 1B). In addition to being an enhancer trap that expresses Gal4 in PS12, the insertion disrupts regulatory sequences and results in a strong iab-7 mutation that, when in trans to Abd-B null mutations, substantially reduces Abd-B expression in the A7, transforms this segment into the A6, and makes the A7 histoblast size and number resemble those of the A6 ( Figure 1C; Figure S1F-G9). We conclude that changes in AbdB expression levels are necessary and sufficient to regulate the differences in cell size and number between the A6 and the A7.
The EGFR pathway regulates the second phase of histoblast development [19]. We have found that the expression of spitz (spi), a ligand of the EGFR pathway present both in histoblasts and LECs [19], and of argos, a target of the pathway [23], are reduced in the male A7 as compared to that of more anterior nests ( Figure 1D-D0; Figure S1I, I9; the reduction is weakly detected in some cases). As expected, this different spi expression depends on Abd-B ( Figure 1E-E0). The down-regulation of EGFR ligand expression seems to be important because forcing the expression of the unprocessed form of Spi (Spi.m) [24] ( Figure 1F; Figure S1J), of an activated form of Ras (Ras V12 ) [25] (Fig, S1K), or of another EGFR ligand, vein [15], allows the formation of a reduced A7 segment (compare with a MD761 UAS-y + /+ male in Figure 1G). Further, the transformation of A7 into A4 observed in MD761-Gal4 UAS-Abd-BRNAi flies (Figure1H) is substantially reduced if we co-express a dominant negative form of the Epidermal growth factor receptor [26] (Figure S1L), a dominant negative form of Raf, a protein that transduces the signal [27,28] (Figure 1I), or the wildtype Argos protein, which inhibits the pathway [23] ( Figure  S1M). We also observed in these mutants changes in histoblast cell number: thus, an increase or a reduction in activity of the EGFR pathway in the A7 augments or diminishes, respectively, histoblast number at about 22-24 h APF ( Figure 1K, K9, L, L9, the wildtype in 1J, J9). In a similar way, if Abd-B expression is reduced, the number of A7 histoblasts increases to resemble that of the A6 ( Figure 1M, M9), and this increase is partially reverted if the EGFR pathway is down-regulated ( Figure 1N, N9). The difference A7 size in these two genotypes is observed later in development, after full expansion of the nests ( Figure 1O, P). All these results suggest that high levels of AbdB down-regulate EGFR activity in the A7 and that this regulation probably impinges in the number of A7 cells and in A7 size after full histoblast expansion.

Male A7 histoblasts are extruded through the epithelium and die
To study why the male A7 histoblasts, although reduced in number, do not form an adult A7 segment, we made time-lapse movies of the posterior abdomen marking posterior compartments with en-Gal4 and nuclei with His2A-RFP. Although it has been shown that some A7a histoblasts show de novo en expression in pupa [15], this change is unlikely to alter the general effects we have seen: at ,25-35 hours APF, we observed the apparent progressive disappearance of the A8 segment, the one abutting the rotating

Author Summary
Many species display sexually dimorphic characters in specific regions of their body. In Drosophila melanogaster, a striking difference between males and females is the development of the seventh abdominal segment (A7), absent in males. We have found that in the first 30 h of pupal development, proliferation in the male A7 is reduced as compared to that of other abdominal segments, resulting in a small primordium. The Epidermal growth factor receptor pathway, which is in part responsible for this reduction, is down-regulated in male A7 cells, and if the activity of the pathway is increased there is a small seventh segment in the adult male. In later stages of pupal development, the remaining cells of the male A7 invaginate and die, and this requires the activity of myosin regulated by the gene extramacrochetae. Extramacrochetae levels of expression are increased in the male, but not female, A7 cells, suggesting that the sex determination pathway regulates this sexual difference (absence or not of the A7) by governing this gene. The Hox gene Abdominal-B, required to specify the posterior abdominal segments, controls both down-regulation of the Epidermal growth factor receptor pathway and extrusion, the latter partly through the regulation of the transcription of doublesex, a key gene in the sex determination pathway.  Figure 2A-A09). This is remarkable, as it suggests that all the LECs of this segment may be extruded without the help of histoblasts (absent in the A8), something that occurs only in a reduced number of LECs from other segments [29]. From ,36 to ,45 h APF we also observed a similar apparent and gradual elimination of the A7 field; as a result of this effect, the A6 cells seem to move backwards, until A6p cells contact with the genital disc (Video S3; Figure 2B-B09). This is also observed with the (p)esg-Gal4 UAS-GFP and neuroglian-GFP (nrg-GFP) [30] markers (Videos S4 and S5; Figure 2C-C09 and Figure S2A). Optical Z-sections of the A7 segment in the former movie show the accumulation of histoblasts underneath the epidermis ( Figure 2D), indicating the A7 histoblasts, like the LECs, undergo delamination (see also below, Videos S9 and S10).
Because of the curvature of the pupal abdomen, a better resolution of the movement and extrusion is observed in Abd-B Fab7-1 homozygous pupae, in which both the A6 and A7 invaginate. In these pupae we observed that the extrusion seemed to be concentrated in two wide regions of cells (left and right) close to the LECs, and where cells show in optical sections a reduced apical size (Video S6 and Figure 2E-E0). This suggests that, similarly to LECs [12], non-muscle myosin may be required for histoblast extrusion. Consistently, the levels of spaghetti-squash, encoding the regulatory light chain of the non-muscle myosin II [31], are elevated at ,35-40 h APF in the male A7 as compared to the A6 ( Figure 2F). Furthermore, expressing a constitutively active form of the myosin binding subunit (MbsN300), a subunit of the phosphatase that inhibits myosin activity [32], we delay extrusion of larval cells [12] and of histoblasts (Video S7 and Figure S2B; the wildtype in Video S8 and Figure S2C). Male adults of the MD761-Gal4 UAS-MbsN300 genotype present a small A7, unpigmented and without bristles ( Figure S2E, compare with the wildtype in Figure S2D). Taken together, the data suggest that A7 histoblasts invaginate like LECs and that myosin II is required for this extrusion.
The extrusion of LECs is accompanied by their death and clearance by macrophages [12,29]. We also observed delamination (Videos S9 and S10; Figure 3A-A00, B-B00) and cell death ( Figure 3C, D) of some histoblasts in the A7 dorsal nests. However, if we inhibit apoptosis by expressing the Diap1 protein, which prevents cell death [33], A7 histoblasts seem to be extruded (Video S11; Figure 3E-E90), though their final elimination takes longer than in the wildtype ( Figure 3F-I). However, cell death, although required for the efficient final elimination of histoblasts, is immaterial as to A7 suppression: the expression of cell deathinhibitors like P35 [34], puckered [35] or Diap1 in the A7 does not prevent the disappearance of this segment [15] (Figure 3J, K and data not shown).
The extramacrochetae gene is required for the extrusion of male A7 histoblasts We have found that males with reduced function in the extramacrocheate (emc) gene, which encodes a HLH protein [36][37][38] with homology to vertebrate ID proteins, develop a small A7 segment ( Figure 4A and Figure  We studied emc expression with an emc-GFP enhancer trap [39] and found that emc is expressed both in LECs and histoblasts. Importantly, male pupae of about 36-42 h APF show an increase in emc-GFP expression in A7 dorsal histoblasts as compared with A6 ones ( Figure 4E). As predicted, this higher expression depends on Abd-B levels ( Figure 4F-F0). Consistently, emc mutations are epistatic over the Abd-B Fab7-1 mutation ( Figure 4G, H) and an increase in Emc can partially suppress the A7 segment produced by Abd-B mutations ( Figure S3N, O). To ascertain the role of emc we made time-lapse movies in ,36-48 h APF emc P5C male pupae and found that the extrusion of the dorsal A7 histoblasts is largely prevented (Video S13, and Figure 4J-J0, compare with the wildtype in Video S12 and Figure 4I-I0), although invagination of larval cells is not greatly disturbed. A similar result is observed in other emc mutant combinations, although a strong reduction in emc levels also affect LECs extrusion (not shown). Collectively, these results strongly suggest that Abd-B promotes suppression of male A7, at least in part, by regulating histoblast extrusion through the control of emc.
Both down-regulation of the EGFR pathway and increased emc expression seem to contribute to the suppression of the male A7 ( Figure S3P-T). Overexpression of spi or Egfr shows mild effects in emc-GFP expression in the male A7 (video S14 and Figure 5A-B9; the wildtype in Figure 4E) but increases the histoblast number ( Figure 1K, K9), so that the size of the A7 segment at about 36-44 h APF pupal stages is bigger than in the wildtype and many histoblasts are not extruded (Videos S14 and S15; Figure 5A-A0, B, B9, C-C00). However, the detailed analysis of these movies suggest that, in addition to an increase in cell number, the strong activation of the EGFR pathway may also reduce extrusion, perhaps due to the slight effect observed in emc-GFP levels.

Interactions between extramacrochetae and wingless in A7 development
A previous study [15] demonstrated that wg is expressed in the female, but not the male, A7 histoblasts, and that ectopic wg develops a small A7 in the male, partially pigmented and without bristles ( Figure S4A). In Abd-B mutants there is ectopic wg in the male A7 [15], and this is important for the formation of the segment since the Abd-B mutant phenotype is partially rescued by diminishing wg activity ( Figure S4B, compare with Figure 1H). To see if emc works in the A7 by regulating wg we looked to wg expression when Emc function is compromised. Wg antibody signal is not detected in the A7 of MD761-Gal4 UAS-GFP UAS-emcRNAi male pupae except, in some of them, for a very faint signal observed in some cells ( Figure S4C). In the reciprocal experiment, however, we note a slight reduction in emc-GFP signal when wg expression is forced in the male A7 histoblasts ( Figure  S4D). This suggests that emc does not prevent A7 development by suppressing wg but that wg may regulate in part emc expression.

A7 development in females
The wildtype female A7 is smaller than the A6 ( Figure 6A). The initial stages of male and female pupal development are similar, including the reduction in cell division rate of histoblasts, although not so strong as in the male [15] ( Figure 6B-B0), and the downregulation of spi expression in the A7 ( Figure 6C, C9). Consistently with a role of the EGFR pathway in controlling the A7 size, we observe that this size increases when we express Spi.m ( Figure 6D) and it is reduced after the expression of a dominant negative form of the Raf protein ( Figure 6E).
A significant difference, however, is seen at later stages. Contrary to what happens in males, the A7 levels of emc-GFP ( Figure 6F) or sqh-GFP ( Figure 6G) at about 35-40 h APF are similar to those observed in the A6, and although some histoblasts seem to be extruded in the central region of the segment ( Figure 6H, H9), the massive effect occurring in males is not observed. However, emc mutant females present a slight but consistent increase in A7 size with respect to the wildtype ( Figure 6I, compare with Figure 6A), perhaps due to the prevention of this extrusion. Consistent with this view, the A7 size increases in MD761-Gal4 UAS-MbsN300 females ( Figure 6J).
Our experiments indicate that emc levels are regulated by Dsx proteins in the A7 and that emc, in turn, regulates sqh: first, increasing emc in the female A7 elevates sqh-GFP levels and suppresses the A7 segment ( Figure 6K, L); second, in XY dsx 1 intersexes, in which neither DsxF nor DsxM isoforms are made and which make a small A7 [40], the amount of emc-GFP in this segment at the time of extrusion seems lower than in the male A7 ( Figure 6M); third, the expression of DsxM in the female A7 variably increases emc-GFP expression ( Figure 6N) and suppresses the A7 ( Figure 6O); finally, the expression of DsxF in the male A7 reduces emc-GFP signal ( Figure 6P) and promotes the development of a segment ( Figure 6Q). Nevertheless, high levels of emc are probably insufficient to determine the suppression of a segment: in pnr-Gal4 UAS-emc male pupae, in which emc expression is increased in the central dorsal region of the whole abdomen, the sqh-GFP signal is not elevated and there is no major extrusion of histoblasts in A6 or anterior segments (Video S16; Figure S5 and data not shown), suggesting that higher Emc levels than those obtained in this combination are required for extrusion and/or pointing to an Abd-B-dependent, emc-independent, contribution to delamination. Taken together, all these results suggest that changes in emc and sqh levels may mediate, at least in part, the activity of Dsx proteins to establish sexual dimorphism in the A7.
Recent results have shown that dsx is only expressed in specific cells throughout development, by and large those that will show sexually dimorphic characters [41][42][43][44][45]. To ascertain the expression of dsx in the posterior abdomen we have used dsx-Gal4 lines [44,45] and found that the expression driven by these lines resembles that of Abd-B, with higher levels in the A7 of male or female pupae ( Figure 6R, S). This suggested that Abd-B may regulate dsx expression and, according with this assumption, we found that down-regulation of Abd-B reduces dsx expression ( Figure 6T). Similar results have been reported recently [46]. Our experiments suggest that changes in DsxF or DsxM levels in the A7, dictated by Abd-B, may mediate Abd-B effects. Consistently, expression of the DsxM protein in and Abd-B mutant background strongly reduces the A7 segment of males or females ( Figure 6U, compare with Figure 1C, and data not shown) and substantially increases emc-GFP expression in males ( Figure 6V). Pupae of this genotype show normal morphogenetic movements in the A7 (Video S17), suggesting the phenotypic rescue is not due to massive cell death.

Discussion
The elimination of a part of an animal body is a major change occurring during morphogenesis and evolution. We have analyzed here the mechanisms required for one such change, the absence of the male seventh abdominal segment. Our study shows that the suppression of this segment involves the interplay between Hox and the sex determining genes, which regulate targets implementing the morphological change. The reduction or suppression of this segment is also a sexually dimorphic feature characteristic of higher Diptera, so the mechanisms shown here may be relevant for the evolution of morphology.
We have shown that in early pupa, during the second phase of cell division, there is a reduction in the number of A7 histoblasts, both in males and females ( [15]; and this report), but stronger in males perhaps because wg is not expressed in the male A7 histoblasts [15]. It has been shown that fewer histoblasts result in a smaller adult segment [47]. Therefore, the reduced number of A7 histoblasts may account in part for the reduced size of the A7 segment in females. The control of the second phase of cell division involves the EGFR pathway [19], and we have found that Abd-B reduces the number of histoblasts in the A7 through downregulation of EGFR activity. If we elevate this activity in the male A7 we observe an increase the number of histoblasts, that many of these cells remain at the surface at the time of extrusion and that a small A7 forms in the adult. It was also previously reported that a small A7 is observed in the male adult when expressing vein, an EGFR ligand [15]. It is possible that the high number of histoblasts obtained when over-expressing elements of the EGFR pathway makes many of them unable to be extruded by a ''titration'' effect, that is, there may be ''too many'' histoblasts for the invagination mechanism to extrude them at the correct time. However, the EGFR pathway may also hinder extrusion since we see lower levels of emc-GFP and also that many histoblasts remain at the surface after high EGFR activation.
At later pupal stages (around 35-40 h APF) there is the extrusion of the male A7 histoblasts. We have observed, however, that a few histoblasts also invaginate in the female A7, suggesting the male intensifies a mechanism present in both sexes. The extrusion requires the activity of emc, and correlates with higher emc expression in the male A7 histoblasts at about the time of extrusion. The invagination of histoblasts superficially resembles that of larval cells [12], and it also requires myosin activity. This would suggest that, due to the higher levels of Abd-B and DsxM, male A7 histoblasts may have adopted a mechanism similar to that used by LECs for their elimination. Recent reports [48,49], however, suggest an alternative mechanism. In these manuscripts the authors demonstrate that an excess of proliferation in the epithelium leads to cell death-independent cell extrusion. Since we have observed that prevention of cell death in the male A7 does not cause the development of an A7 (although delamination is delayed), the mechanism driving extrusion may be more similar to that of an overproliferating epithelium than to that taking place in larval cells.
Our data are consistent with emc increasing the expression of spaghetti-squash to accomplish apical constriction and extrusion. However, high expression of emc may not be sufficient to effectively induce histoblast extrusion, suggesting other genes are required. Besides, a strong reduction of emc leads to a very small and poor differentiated male A7 segment (not shown), reflecting that this gene is required for several cellular functions, among them cell survival [50,51]. Perhaps significantly, emc is also expressed in embryonic tissues preceding invagination of different structures in the embryo [52], suggesting a common requirement for invagination at different developmental stages. We think that emc forms part of complex networks that have, among other cellular functions, that of contributing to the extrusion of A7 histoblasts.
Although regulation of the EGFR pathway and emc are two key events in controlling male A7 development, previous experiments have also shown the contribution of the wingless gene, absent in male A7 but present in male A6 and female A7, in the development of this segment [15]. We have confirmed these results and also shown that a reduction in wg expression can partially suppress the Abd-B mutant phenotype. Absence of wg is probably required to reduce cell proliferation in the male A7 [15] but our data suggest wg may also be needed to maintain high emc levels. Apart from the role of wg, it was also shown that some A7a cells are transformed into A6p cells, thus reducing the number of A7 cells that might contribute to the adult segment [15]. Finally, the expression of bric-a-brac must also be down-regulated in male A7 histoblasts to eliminate this metamere [8]. Thus, this suppression is a complex process using different genes and mechanisms.

Sex determination, Hox information and segment elimination
The suppression of the male A7 depends ultimately on the levels of Abd-B expression. The role of this Hox gene is probably mediated in part by dsx, since Abd-B regulates dsx transcription ( [46]; and this report) and dsx governs, in turn, the expression of genes required for cell proliferation and extrusion (Figure 7). That Hox genes regulate dsx expression has also been demonstrated in the male foreleg [53], suggesting that Hox genes specify the different parts of the body where sexual dimorphism may evolve. The different dsx isoforms (DsxF and DsxM) determine the outcome of this regulation. A significant difference between the activities of these two proteins in the A7 is the regulation of emc levels. In the female, emc expression is similar in the A7 and the A6 and, accordingly, histoblast extrusion in females is small and confined to the central dorsal region, a domain virtually absent in the adult tergite. By contrast, the DsxM isoform increases Emc expression to drive large extrusion of A7 cells and elimination of the segment (Figure 7).
Only the male A7, but not anterior abdominal segments, is eliminated. Therefore, the increase in emc expression, and subsequent events observed in the A7, depends on the higher Abd-B expression in the A7 in relation to the A6. Several Hox loci, like Sex combs reduced, Ultrabithorax or Abd-B are haplo-insufficient, and relatively small differences in the amount of some of these Hox proteins can drive major phenotypic changes [54][55][56], suggesting some downstream genes can sense these slight differences and implement major changes in morphology.
Previous studies have shown the cooperation of Abd-B and the sex determination pathway in controlling the pigmentation of the posterior abdomen [8,9]. We think that Abd-B plays a dual role in regulating the morphology of the posterior abdomen. First, it regulates dsx expression, thus allowing the possibility to develop sexually dimorphic characters; second, it cooperates with Dsx proteins in establishing pattern (Figure 7). Part of the effect implemented by Abd-B may be mediated by the levels of expression of dsx (distinguishing male A6 from male A7), and from the nature of the Dsx proteins (male and female ones). Although there is no conclusive evidence that the different levels of dsx in the A6 and A7 play a role in development, we note that this difference correlates with that of Abd-B (and depends on it), that high levels of DsxM are sufficient to increase emc-GFP in the A7 of females and eliminate this segment, and that these same high levels similarly increase emc-GFP and partially rescue the Abd-B mutant phenotype in males. Hox genes, therefore, may provide a spatial cue along the anteroposterior axis to activate dsx transcription and allow the formation of sexually dimorphic characters, but they may also cooperate with Dsx proteins to determine different morphologies. This double control by Hox genes may apply to all the sexually dimorphic characters and be also a major force in evolution.

Immunohistochemistry
The primary antibodies used are: mouse anti-Abd-B at a 1:100 dilution ( [62]; and Developmental Studies Hybridoma Bank, University of Iowa), and mouse and rabbit anti-ß-galactosidase at 1:2000 (Cappel). Secondary antibodies were conjugated antimouse or anti-rabbit Fluor 488, 555 or 647 (Alexa) used at a 1:200 dilution. Topro (TO-PRO-3; Molecular probes) was used to mark nuclei. Immunostaining and sample preparation were done according to standard methods. Pupal cuticle staining was performed as described [12] with small variations. White prepupa were transferred to empty vials and kept at 25uC for staging. Whole pupae were then bisected with a razor blade, cleaned with PBS and fixed for 90 minutes in 4% paraformaldehyde at 4uC, rinsed four times 615 minutes in PBT-Triton (0.1% Triton X-100, 1% BSA in PBS) and blocked for at least 1 hour using PBT-BSA (1% bovine serum albumin (BSA) in PBT). Antibodies treatment and mounting were done following standard procedures.

In vivo imaging and image analysis
Leica TCS SPE and Zeiss LSM700 confocal microscopes were used to capture both still images and time-lapse movies. All the confocal images are maximum intensity projections. Staging of the pupae was performed as described [12]. APF stands for hours after puparium formation, taking the eversion of anterior spiracles in white prepupae as a reference. Male (XY) pupae of the genotype X B S Y dsx 1 /dsx 1 were distinguished from the XX siblings by the B S mutation. The use of different setting conditions in the capturing the different movies makes to see the rotation of the genitalia look normal (clockwise) or inverted (anti-clockwise). All the movies were captured at 10 minutes intervals keeping the laser intensity at a minimum to avoid damaging the pupae. Unless specified, all the images correspond to z-stacks with slides taken at an optimum distance to get the whole structure 3D reconstruction of z-stacks, and mounting of time-lapse movies in AVI format was performed with Leica Confocal Software (LAS AF Lite) or Zeiss ZEN2009 software. ImageJ (NIH Image) and Photoshop 7.0 (Adobe Corporation) were used for data processing, cell counting and measurement of signal intensity.

Adult cuticle preparations
Flies were kept in a mixture of ethanol: glycerol (3:1), dissected, macerated in 10% KOH-at 90uC for three minutes, washed with MD761-Gal4/Abd-B M1 ), it almost completely rescues the transformation induced by the loss of Abd-B (compare with Figure 1C). 5 males of this genotype show this phenotype whereas 4 other males present a small A7 segment, sometimes in only one side. Cross made at 17uC. (V) In a male pupa of the UAS-DsxM/+; emc-GFP MD761-Gal4/Abd-B M1 genotype the levels of emc-GFP expression in the A7 are high. doi:10.1371/journal.pgen.1002874.g006 Video S8 Disappearance of the A7 segment in His2A-RFP MD761-Gal4 UAS-GFP males. The timing of the movie is like that of the previous one. Note that, different from it, larval cells and histoblasts are being extruded. His2A-RFP, marking nuclei, is in red, and the A7 segment is labeled in green. Note the rotation of the genitalia. See also Figure S2C. (AVI) Video S9 Delamination of histoblasts (I). We show a movie from about 38-44 h APF in a (p)esg-Gal4 UAS-GFP male pupa. Left and right histoblast nests from the A7 meet at the dorsal midline and at the same time the histoblasts delaminate. See also Figure 3A-A00.

(AVI)
Video S10 Delamination of histoblasts (II). The movie, from ,42 to 50 h APF, represents the delamination of A7 histoblasts, marked in green in a (p)esg-Gal4 UAS-GFP male pupa. Note that at the beginning of the movie the width of the epithelium in the A7 region is bigger than at the end of the movie (after delamination). Unlike in the rest of the movies, Z stacks were collected every 5 min. instead of 10 min. to better visualize the delamination of cells. See also Figure  Video S16 Ectopic emc is not sufficient to extrude histoblasts of segments anterior to the A7. The movie has been obtained in a UAS-emc/UAS-GFP; pnr-Gal4/+ male of ,36-50 h APF. See that the histoblasts of the A6 and A5 segments remain at the surface. See also Figure S5. (AVI) Video S17 The expression of DsxM increases emc-GFP expression in an Abd-B mutant background. The genotype of the male pupa is UAS-DsxM/+; emc-GFP MD761-Gal4/Abd-B M1 . There is increase of emc-GFP signal in the A7 even though the pupa is mutant for Abd-B. See normal cell divisions and cell movements. See also Figure 6V. (AVI)