The Spalt transcription factors regulate cell proliferation, survival and epithelial integrity downstream of the Decapentaplegic signalling pathway

The Drosophila spalt complex contains two genes, spalt major (salm) and spalt related (salr), both encoding nuclear proteins which main feature is the presence of several Zn-finger domains of the C2H2 class (de Celis and Barrio, 2009). These proteins are conserved through evolution, and play important roles during animal development in a variety of processes such as cell fate specification, pattern formation and organogenesis (Sweetman and Münsterberg, 2006; de Celis and Barrio, 2009). Two of the four human orthologues of Sal are associated to the genetic diseases Townes–Brocks syndrome (Sall1) and Okihiro syndrome (Sall4) (Domingos et al., 2004; de Celis and Barrio, 2009). Sall4 is also involved in embryonic stem cells maintenance interacting with Nanog (Wu et al., 2006). The molecular characteristics of Sal proteins, and the multitude of functional requirements identified for sal genes in different organisms, suggest that these proteins act as transcriptional regulators in a context-dependent manner.

The salm and salr genes have been extensively characterised during embryonic and imaginal disc development (Kühnlein et al., 1994; Kühnlein and Schuh, 1996; de Celis and Barrio, 2000; Rusten et al., 2001; Cantera et al., 2002). Thus, sal function is required for the specification of several cell types including the oenocytes (Gould et al., 2001), the R7 and R8 photoreceptors (Mollereau et al., 2001) and sensory organ precursors of the thorax (de Celis et al., 1999), and they also participate in the formation of the tracheal system, the migration of the tracheal dorsal trunk cells (Kühnlein and Schuh, 1996), neural development (Cantera et al., 2002), Johnston organ formation (Dong et al., 2003) and wing imaginal disc development (de Celis et al., 1996). In this last process, the Decapentaplegic (Dpp) signalling pathway regulates the expression of sal genes in the wing blade region of the disc, but it is not yet clear to what extent these genes mediate the function of this pathway.

The wing imaginal disc is an epithelial tissue that grows by cell proliferation during larval development and differentiates the wing and half of the thorax during pupal development (de Celis, 2003). The growth of the epithelium is linked to the progressive specification of spatial territories with different genetic identities and cell fates. Several signalling pathways play a fundamental role in these processes, in part by regulating the expression of transcription factors. Of paramount importance for the growth and patterning of the wing disc is the Dpp signalling pathway (Affolter and Basler, 2007). The ligand of the pathway, Dpp, is expressed in a narrow stripe of anterior cells abutting the anterior–posterior compartment boundary of the disc. Dpp secreted from these cells binds and activates its receptor complex, formed by the transmembrane kinases Thickvein (Tkv) and Punt (Put), leading to the activation of the pathway in a broad stripe of anterior and posterior cells occupying most of the presumptive central region of the wing (Affolter and Basler, 2007). The domain of Dpp signalling accumulates phosphorylated Mad (PMad), a direct target of the Tkv/Put kinase (Affolter and Basler, 2007). Once phosphorylated, PMad forms a complex with Medea (Med) that enters the nucleus and regulates the expression of target genes. Mad and Med are conserved through evolution, being the orthologues of vertebrate Smad3 and Smad4 proteins, respectively (Massagué and Wotton, 2000). Several targets and additional components of the transcriptional regulation events triggered by Dpp signalling have being identified in Drosophila (Affolter and Basler, 2007), but it is not clear which of the different functions of the pathway are mediated by these known targets. These functions include the promotion of cell proliferation throughout the wing blade primordium, the patterning and differentiation of the longitudinal and transverse veins during larval and pupal development and the maintenance of epithelial integrity in the wing disc epithelium (Affolter and Basler, 2007). Because the sal genes are downstream components of the Dpp pathway in the wing blade (Lecuit et al., 1996; Nellen et al., 1996), and are also required for wing patterning (de Celis et al., 1996), it is likely that they mediate some of the requirements of Dpp during the development of the wing blade.

In this work we have dissected the functions of Sal proteins during the development of the wing blade, aiming to relate its requirements to the known functions of Dpp signalling. We find that Salm/Salr are required for the proliferation of cells located in its domain of expression, promoting the transition between G2 and Mitosis. Consequently, wing blade cells deprived of sal function accumulate in the G2 phase of the cell cycle in the developing disc. We also find that loss of salm/salr in the wing disc results in apoptosis, indicating an unsuspected pro-survival function of these genes in the central region of the wing blade. In addition to a requirement for cell proliferation and survival, salm/salr functions are also needed to maintain epithelial integrity in the wing disc and pupal wing. Thus, the boundaries between sal-expressing and non-expressing cells lead to the formation of grooves in the disc epithelium, and subsequently during pupal development to the extrusion of mutant cells. Finally, we find that loss of salm/salr and gain of salm expression can rescue some of the effects caused by excess and loss of dpp function, respectively, indicating that Sal proteins mediate in part the function of Dpp signalling during wing development.

We used the Gal4 lines dpp-Gal4, en-Gal4 and sal^EPv-Gal4 (Cruz et al., 2009), the tub-Gal80^ts line, and the UAS lines UAS-GFP (Ito et al., 1997), UAS-salm (de Celis et al., 1996), UAS-brk (Jaźwińska et al., 1999), UAS-tkv^QD (Nellen et al., 1996), UAS-dppGFP (Teleman and Cohen, 2000), UAS-puc (Martín-Blanco et al., 1998), UAS-stg and UAS-dicer2 (Dietzl et al., 2007). The expression of dpp-Gal4 is restricted to the anterior–posterior compartment boundary of the wing imaginal disc. The expression of en-Gal4 is restricted to the posterior compartment of all imaginal discs, and the expression of sal^EPv-Gal4 occurs in the central region of wing imaginal disc between the vein L2 and intervein L4–L5. We also used the following UAS lines to express interference RNA: UAS-salm-i (ID 3029 VDRC), UAS-salr-i (ID 28386 VDRC), UAS-Mad-i (ID 12399-R2 NIG-FLY) and UAS-CycA-i (ID 32421 VDRC), and the loss of function alleles Df(2L)32FP5 (Barrio et al., 1999) and tkv^a12 (Burke and Basler, 1996). Unless otherwise stated, crosses were done at 25°C. Lines not described in the text can be found in FlyBase (Gelbart et al., 1997).

Marked clones were generated by FLP-mediated recombination (Xu and Rubin, 1993) by 1 hour heat shock at 37°C 48–72 h and 72–96 h after egg-laying. Clones were visualised in third instar larvae by the absence of β-gal expression or by the presence or absence of GFP expression, and in the adult wing by the cell marker forked (f). Clones were induced in larvae of the following genotypes:

1. hsFLP1.22/+; FRT40 tkv^a12/FRT40 P[armlacZ]

2. hsFLP1.22/+; FRT40 Df(2L)32FP5/FRT40 P[armlacZ]

3. y w hsFLP1.22 P[tub-Gal4] UAS-GFP/+; FRT40A P[tub-Gal80]/FRT40 UbiGFP; UAS-salm/+

4. y w hsFLP1.22 P[tub-Gal4] UAS-GFP/+; FRT40A P[tub-Gal80]/FRT40 tkv^a12; UAS-salm/+

5. y w hsFLP1.22 P[tub-Gal4] UAS-GFP/+; FRT40A P[tub-Gal80]/FRT40A Df(2L)32FP5; UAS-tkv^Q253D/+

6. f^36a hsFLP1.22; FRT40A Df(2L)32FP5/FRT40A ck P[f⁺]32F.

sal Minute+ clones were induced by X-rays as described by de Celis et al. (de Celis et al., 1996) in larvae of f^36a; Df(2L)32FP5/P[f⁺]32F M(2)z genotype. Homozygous sal mutant cells were labelled with the cell marker forked (f). We quantified the anterior or posterior area of pairs of wings from the same fly bearing either anterior or posterior sal⁻Minute⁺ clones, and calculated the average value of the ratios between the sizes of mutant anterior compartment and control anterior compartment of the contralateral wing, and between the sizes of mutant posterior compartment and control posterior compartment of the contralateral wing.

We quantified the ratio between the number of cells in mitosis in the central region of the wing blade, detected by the expression of Phospho-Histone3 (PH3), and the size of this region, and calculated a mitotic index as the average value of these ratios. The images used to make the quantifications were projections of at least 10 confocal planes including the entire height of the wing blade epithelium. We did these measures in third instar wing discs of the following genotypes: UAS-dicer2/+; sal^EPv-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/+ and UAS-dicer2/+; sal^EPv-Gal4 UAS-GFP/+. The expression of GFP was used to label the central region.

We incubated wing imaginal discs of UAS-dicer2/+; sal^EPv-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/+ and UASdicer2/+; sal^EPv-Gal4 UAS-GFP/+ (control discs) in 300 µl of trypsin solution (Trypsin–EDTA Sigma T4299) and 0.5 µl Hoescht (Hoescht33342, Trihydrochloride Trihydrate H3570, Molecular Probes^TM) at 28°C during 40 minutes. We stopped the trypsin reaction with 200 µl of 1% FCS (Fetal Bovine Serum Sigma^TM 9665). The cells were suspended in 300 µl of 1% FCS and the cell cycle profiles of GFP positive and GFP negative cells was quantified with a FACS Vantage 2 (Becton Dickinson^TM). The cell cycle profile of 5 independent experiments for each genotype was analysed using FloJow 7.5. We quantified for each genotype the G1/G2 ratio in the central region (GFP positive) and in the rest of the wing imaginal disc (GFP negative).

We used rabbit anti-Salm and rat anti-Salm (de Celis et al., 1999), rabbit anti-activated Cas3 (Cell Signalling Technology) and anti-Phosphorylated Histone 3 (PH3; Cell Signalling Technology), mouse anti-βGalactosidase (Promega), TO-PRO (Invitrogen) and Phalloidine (Sigma). From the Hybridoma bank at Iowa University we used monoclonals anti-Wg, anti-FasIII, anti-Dlg and anti-Dl. Primary antibodies were used at a 1:50 dilution, except anti-Dl (1:5). Secondary antibodies were from Jackson Immunological Laboratories (used at 1/200 dilution). Imaginal wing discs and wing pupal were dissected, fixed and stained as previously described by de Celis and Bray (de Celis and Bray, 1997). Confocal images were captured using a LSM510 confocal microscope. In situ hybridization with stg RNA probes was carried out as described by de Celis and Bray (de Celis and Bray, 1997).

We analysed at least 10 wings from females of each genotype. Wing size, imaginal disc size, cell size, cell number data and ratios between number of cells were expressed as means ± standard error of the mean (SEM) and were compared using a T-test. All data were collected and processed in Microsoft Excel (Microsoft Inc.). We consider a significant p-value lower than 0.005.

The expression of salm and salr is regulated by Dpp signalling in the wing pouch and occupies a broad stripe of cells centred along the anterior–posterior compartment boundary (Fig. 1H,H′). This territory corresponds in the wing to a region including the vein L2 and extending to the L4/L5 intervein (Fig. 1A,J). The sal^EPv-Gal4 driver reproduces the domain of sal expression in the wing blade region of the imaginal disc (Cruz et al., 2009) (Fig. 1I, compare with Fig. 1H,H′), and this domain can also be visualised in pupal wings (Fig. 1J). The reduction of only salm or salr expression in the domain of sal^EPv-Gal4 has a modest phenotype in adult wings, which are slightly reduced in size and display ectopic vein stretches in the region between the veins L2 and L3 (Fig. 1B,C,E; UAS-dicer2/+; sal^EPv-Gal4 UAS-GFP/UAS-salm-i and UAS-dicer2/+; sal^EPv-Gal4 UAS-GFP/+; UAS-salr-i/+). As expected, wing discs expressing salm and salr RNAi show very much reduced levels of Salm (Fig. 1K′) and Salr proteins (not shown), respectively. The over-expression of salm in its normal domain of expression (sal^EPv-Gal4/+; UAS-salm/+) results in a phenotype similar to the loss of only salm or salr (Fig. 1G). This is in contrast to the consequences of ectopic and generalised expression of either salm or salr, which results in very small wings with severe defects in the pattern of veins (de Celis et al., 1996). When both salm and salr expression are reduced, the wing shows a phenotype which strength depends on the particular genetic combination used and temperature. The stronger version of this phenotype consists in wings losing most of its central region and the veins L2 and L4 (Fig. 1D,E; UAS-dicer2/+; sal^EPv-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/+). The corresponding discs show a domain of sal^EPv-Gal4 expression reduced in size (Fig. 1K), and this reduction is even stronger in pupal wings (Fig. 1L, compare with Fig. 1J). The extreme phenotype caused by a reduction in salm and salr expression is also observed in wings homozygous for a genetic deficiency that removes the salm and salr genes (Fig. 1F; 638-Gal4/+; FRT40 Df(2L)32FP5/FRT40 M(2)z; UAS-FLP/+). In this manner, the central region of the wing fails to grow in the absence of both salm and salr, but is very much insensitive to the increase in the expression of these proteins, as far as this is restricted to the normal domain of expression.

We followed the development of the central region of the wing blade in wing discs and pupal wings of UAS-dicer2/+; sal^EPv-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/+ genotype. The sal domain of expression in third instar discs, labelled by GFP, is smaller than normal in mutant discs, and in general the appearance of the epithelium is normal (Fig. 2A–A″, compare with Fig. 1I and Fig. 2C–C″). We also observed the presence of activated-Cas3 positive cells located in the basal part of the epithelium of the Sal domain (Fig. 2B–B″), and a variable formation of epithelial folds (Fig. 2B′,B″) that is never observed in control discs (Fig. 2C–C″). The expression of FasIII, which is mostly localised along the baso-lateral side of the epithelium in wild type discs, is very much reduced in sal mutant cells compared to wild type cells (Fig. 2D,E). The mutant sal^EPv-Gal4 expressing cells are also present during pupal development (24–40 h APF), although at these stages they have an abnormal cell size, being much larger than normal cells (Fig. 2F,G). We also found that the integrity of the epithelium is compromised in 24–30 h APF pupae (Fig. 2H–K), and that the contacts between the dorsal and ventral wing surfaces appear abnormal (Fig. 2H,I, Fig. 2H′,I′). We observed groups of cells that seem extruded from the epithelium (Fig. 2I′), but detected very few apoptotic cells (see below). In late pupa 36–40 h APF, the central domain of the wing blade can form abnormal blisters (Fig. 2K). In all cases we find a large number of GFP-negative cells located between the dorsal and ventral wing surfaces, which likely correspond to circulating haemocytes (Fig. 2I–I‴, Fig. 2K–K‴). These cells are normally localised in the veins (Fig. 2H,H′), but in sal mutant pupa they also appear intercalated within the epithelium (Fig. 2I–I‴, Fig. 2K–K‴). In this manner, the domain of sal expression grows to a smaller than normal size during imaginal development, and some cells are extruded from the epithelium. The extrusion of cells is more dramatic during pupal development, and after this stage only a small fraction of the central domain remains in the adult wing.

The small size of the mutant sal domain of expression suggests a role for salm and salr in the regulation of cell proliferation and/or cell survival. Clonal analysis of sal mutant cells confirms a cell-autonomous requirement for these genes in the central region, where sal mutant clones are rounder and smaller than their corresponding twins in both wing discs and in adult wings, and a considerable fraction of wild type twins (70%) appear without associated salm/salr mutant clone (de Celis et al., 1996; Milán et al., 2002) (Fig. 3A,B; supplementary material Fig. S1B–D). In addition, the size of anterior or posterior compartments formed by salm/salr mutant cells is strongly reduced, whereas the opposite compartment, formed by wild type cells, is not reduced in size (Fig. 3C–F; supplementary material Fig. S1). To identify which aspects of cell proliferation are affected in salm/salr mutant cells, we first measured the fraction of mitotic cells present in the sal^EPv-Gal4 domain of expression of wing discs growing with reduced levels of salm and salr. We found that the fraction of mitotic cells is strongly reduced in mutant (UAS-dicer2/+; sal^EPv-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/+; Fig. 3H,H′,I) compared to wild type discs (UAS-dicer2/+; sal^EPv-Gal4 UAS-GFP/+; Fig. 3G,G′,I), indicating that entering into mitosis is affected in cells with reduced expression of sal located in the central domain of the wing blade. We also measured the fraction of cells with reduced levels of salm/salr in the different phases of the cell cycle, and found a significant accumulation of salm/salr knockdown cells in G2 compared to wild type cells of the same disc and with wild type cells of control discs (Fig. 3J–K″). In this manner, we suggest that Sal promotes the progression from G2 to Mitosis, and consequently, cells in the sal domain of expression growing with reduced levels of sal expression accumulate in G2, causing a depletion of mitotic cells. The activity of the phosphatase String (stg, orthologous of cdc25) is a key determinant of the G2 to M transition in the wing disc (Edgar and O'Farrell, 1989; O'Farrell et al., 1989). We studied the expression of stg in sal mutant discs by in situ hybridization, and found a reduction of stg expression in the central domain of sal mutant discs (UAS-dicer2/+; sal^EPv-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/+; Fig. 3L′,M) compared to control discs (sal^EPv-Gal4/+; Fig. 3L,M). However, the pattern of stg expression is very variable from disc to disc, both in wild type and in sal mutant discs (data not shown). When we quantified the ratio of stg-expressing cells relative to the area of the disc we found a 15% reduction in sal knockdown discs, but the dispersion of the mean was very high (Fig. 3M). Ectopic expression of stg in the central domain of the wing disc (sal^EPv-Gal4 UAS-salm-i/+; UAS-salr-i/UAS-stg; Fig. 3N′,O) did not rescue the size of wings growing with reduced levels of sal expression (sal^EPv-Gal4 UAS-salm-i/+; UAS-salr-i/UAS-GFP; Fig. 3N,O). In this manner, we interpret that the accumulation of sal mutant cells in G2 is not caused by defects in stg expression, and that the variable reduction in the number of cells expressing stg might be related to the stoppage in G2 of sal mutant cells before they initiate stg expression.

The smaller than normal size of sal mutant clones, and the reduction of the size of the sal mutant central domain of the wing blade can also result from cell death, which is a rare occurrence in wild type discs (Milán et al., 1997). We found small numbers of wing cells expressing activated Cas3 in central domain of the wing blade of salm/salr knockdown wing discs (UAS-dicer2/+; sal^EPv-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/+; Fig. 4A). These cells were located mostly in the periphery of the central domain, and in general they are positioned in the basal side of the epithelium (Fig. 4A–A‴). In contrast, cell death is very rare during pupal development (Fig. 4B–B‴), indicating that the observed epithelial disruption occurring at this stage is not related to apoptosis. We also looked at cell death in a situation where sal mutant and non-mutant cells are confronted along the central domain of sal expression. In wing discs where the expression of salm/salr is reduced in the posterior compartment (en-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/+) we observed a robust induction of cell death located mainly along the boundary confronting sal mutant and not mutant cells (Fig. 4C,C′). However, we did not observe any significant rescue of the sal mutant phenotype in genetic backgrounds in which cell death is prevented, suggesting that the contribution of cell death to the sal phenotype is minor (supplementary material Fig. S2). The activity of the JNK signalling pathway is related to cell death in the wing disc. We looked at the reported of JNK activity puckered (puc) (Martín-Blanco et al., 1998), and found a robust and cell autonomous induction of puc expression in sal mutant discs (en-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/puc-lacZ (Fig. 4D,D′) and sal^EPv-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/puc-lacZ (Fig. 4F,F′)). The suppression of JNK activation in the salm/salr knockdown genetic background does not correct the sal mutant phenotype (sal^EPv-Gal4/UAS-salm-i; UAS-salr-i/UAS-puc (Fig. 4H–I′), compare with sal^EPv-Gal4/UAS-salm-i; UAS-salr-i/+ (Fig. 4E–G′)). In fact, suppressing JNK signalling by puc over-expression aggravates the phenotype of sal mutant wings (compare Fig. 4E, Fig. 4H), perhaps because mis-specified cells are not removed from the epithelium during wing disc development.

The shape of sal mutant clones in the central region of the wing blade is rounded, suggesting that sal mutant cells tend to minimize their contacts with wild type cells (Fig. 3B, Fig. 5A) (see also Milán et al., 2002). In addition, we also observed that the boundary of these clones form grooves separating mutant and normal cells (Fig. 5A′,A″). The formation of grooves separating sal mutant and non-mutant cells is always induced when salm and salr expression is reduced within its normal domain of expression. For example, reduction of salm/salr in the domain of dpp expression (UAS-salm-i/+; dpp-Gal4 UAS-GFP/UAS-salr-i), generates a strong groove that also affects the neighbouring wild type cells (Fig. 5B–C″). Similarly, knockdown of salm/salr in the posterior compartment (en-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/+) also generates a deep groove separating anterior and posterior cells (Fig. 5D–D″). We followed the temporal sequence of groove formation in discs of en-Gal4 UAS-GFP/UAS-salm-i; UAS-salr-i/tub-Gal80^ts genotype. We grow larvae of this genotype at 17°C, preventing the action of the Gal4 protein, and transferred them to 29°C at the beginning of the third larval instar. We observed that Gal4 activity, revealed by the expression of GFP, becomes apparent at 6 hours after the temperature shift (Fig. 5E,E′). At this time the expression of Salm is only moderately reduced, and the epithelium in normal (Fig. 5E′,E″). The expression of GFP together with the loss of Salm protein is very robust 24 h after the temperature shift (Fig. 5F). In these discs, the epithelium still maintains its continuity without any indication of groove formation (Fig. 5F″). Groove formation becomes apparent in discs 48 hours after the temperature shift (Fig. 5G–G‴). At this time, the posterior compartment is reduced in size and is separated from the anterior compartment by a continuous epithelial groove (Fig. 5G–G‴). The formation of grooves is not the result of a discrepancy in size between the posterior and anterior compartments, because when cell proliferation is strongly reduced in posterior cells (en-Gal4 UAS-GFP/UAS-CycA-i; tub-Gal80^ts/+), the boundary between the anterior and posterior compartments is smooth (Fig. 5H–H‴).

From the previous analysis, we conclude that Salm/Salr function is required for cell proliferation and survival, and for the maintenance of epithelial integrity, perhaps determining cell affinities in the central region of the wing disc. These requirements are similar to those reported for different components of the Dpp pathway, and consequently we analysed whether Salm/Salr not only are transcriptional targets of Dpp, but whether they also mediate the functions of this pathway during wing development. We first aimed to rescue the consequences of loss of dpp function by ectopic expression of Salm. The expression of interference RNA against Mad, the transducer of the Dpp pathway, in the central region of the wing disc (sal^EPv-Gal4 UAS-GFP/+; UAS-Mad-i) results in smaller than normal wings that fail to differentiate the veins L2, L3 and L4 (Fig. 6A,B). The corresponding wing imaginal discs show, as expected, very much reduced levels of Salm protein (Fig. 6F, compare with Fig. 6D), and also the loss of Delta expression in the presumptive veins L3 and L4 (Fig. 6G, compare with Fig. 6E). The expression of Salm in this background of reduced Mad (sal^EPv-Gal4/+; UAS-Mad-i/UAS-salm) rescues to a large extent the wing phenotype and the expression of Delta (Fig. 6C,I), despite a patchy accumulation of Sal protein (Fig. 6H). Over-expression of Brinker (Brk) in the central domain of the wing disc (sal^EPv-Gal4/+; UAS-brk/+) generates a strong reduction of wing size and the loss of L2 and L4 veins (Fig. 6J). In the corresponding discs the expression of Salm is entirely lost (Fig. 6K). The presence of Salm in sal^EPv-Gal4/+; UAS-brk/UAS-salm discs does not rescue the growth and pattern defects caused by brk over-expression (Fig. 6L). However, we find that Salm expression is barely detected in these mutant discs (Fig. 6M), most likely because the sal^EPv-Gal4 driver is repressed upon Brk over-expression. We aimed to prevent the effects of Brk and Mad on the regulation of the sal^EPv-Gal4 driver in discs of sal^EPv-Gal4/UAS-salm; tub<FRT-GFP-FRT>Gal4 UAS-FLP/UAS-Mad-i and sal^EPv-Gal4/UAS-salm; tub<FRT-GFP-FRT>Gal4 UAS-FLP/UAS-brk genotypes. In these discs, the expression of Gal4 in the central region of the wing disc is first initiated by the sal^EPv-Gal4 driver and then clonally sustained by the tub-Gal4 driver, once the FRT cassette is removed by FLP-mediated recombination. Unfortunately, we found that the over-expression of salm using the tub-FRT-Gal4 driver results in the generation of clones of cells expressing high levels of salm, leading to the formation of small wings with a severe phenotype of epithelial extrusion (data not shown). The behaviour of clones of cells over-expressing Salm prevented us from checking its ability to rescue the brk and Mad-i phenotypes.

We reason that some consequences of Dpp hyper-activation might be due to the ectopic or increased expression of the sal genes. For example, wing discs expressing the constitutively activated version of the receptor Tkv in the central region of the wing (sal^EPv-Gal4/+; UAS-tkv^QD/UAS-GFP) show higher than normal levels of Sal and a strong disruption of the epithelium compared to wild type discs (Fig. 7A′–B‴). The adult wings of this genotype display a non-autonomous loss of anterior and posterior wing tissue (Fig. 7B). Loss of salm/salr in the Tkv^QD background (sal^EPv-Gal4/UAS-salm-i; UAS-tkv^QD/UAS-salr-i) entirely rescues all components of the Tkv^QD phenotype (Fig. 7C), and the resulting wing discs and adult wing show the characteristic defects of salm/salr loss of function (Fig. 7C′–C‴). Ectopic expression of Dpp in the central region of the wing blade (sal^EPv-Gal4 UAS-dpp-GFP) results in wings that differentiate extra-vein tissue in the L2/L3 and L4/L5 interveins (Fig. 7D). Loss of salm/salr in this background (sal^EPv-Gal4 UAS-dpp-GFP/UAS-salm-i; UAS-salr-i/+) results in smaller wings that also differentiate extra-veins in the same wing regions (Fig. 7F, compare with Fig. 7D,E). In this manner, it seems that the promotion of vein differentiation caused by Dpp does not require Sal function, whereas the growth-promoting function of Sal is not rescued by ectopic Dpp signalling.

We also studied the contribution of Sal to the cell proliferation effects caused by loss or gain of Tkv function. The behaviour of Tkv^QD-expressing clones depends on the position of these clones along the anterior–posterior axes (Burke and Basler, 1996; Rogulja and Irvine, 2005). Thus, clones in the periphery of the wing blade cause extreme overgrowths, due to the repression of brk transcription (Schwank et al., 2008) (Fig. 8A–B′). Clones located in the central region of the wing blade also behave abnormally, as they are also larger than normal and are surrounded by epithelial grooves (Fig. 8A–B′). The loss of sal function in cells expressing Tkv^QD has very different effects depending on their location in the wing blade. Thus Tkv^QD/Dfsal clones in the central region of the wing blade are smaller than normal and behave as Dfsal clones (Fig. 8C–D′). In contrast, clones in the periphery of the disc still cause extreme overgrowth, indicating that Sal only mediates the functions of Tkv in the central region of the wing, but does not contribute to the overgrowth caused by ectopic pathway activation in the peripheral regions. Complementary, we also observed a rescue of the cell lethality of tkv mutant clones in the central region of the wing blade. Thus, tkv^a12 clones fail to grow in the wing blade (Burke and Basler, 1996; data not shown), but when these cells express Salm they can grow in the centre of the wing, forming clones that tend to extrude from the epithelium (Fig. 8G–H′). As mentioned before, clones of cells over-expressing sal also cause epithelial defects and clone extrusion (Fig. 8E–F′).

In this work we describe the requirements of the Sal transcription factors during the development of the wing blade. This description extends the analysis of Sal activity in regulating pattern formation and cell affinities (de Celis and Barrio, 2000; Milán et al., 2002), and identifies additional functions in cell survival, division and maintenance of epithelial integrity. We also show that some of the effects of Dpp signalling during the growth and patterning of the wing disc are mediated by Sal activity. The salm and salr genes were among the first identified transcriptional targets of Dpp signalling (de Celis et al., 1996), and given the key roles this pathway plays during imaginal development, it is important to understand whether its functions are mediated by the Salm and Salr proteins.

The analysis of sal mutant cells, and of wing discs growing with reduced levels of Sal proteins, identifies several requirements for these genes. First, Sal expression in the central domain of the wing blade seems to confer these cells with some affinity label that makes them different from other cells non-expressing the genes. This is reflected in the shape of sal mutant clones, which are rounder than normal (Milán et al., 2002), and in the formation of epithelial folds at the place of confrontation between sal-expressing and sal-non-expressing cells located in the central region of the wing. The nature of this affinity label seems to be quantitative, because clones of cells over-expressing sal also manifest an aberrant behaviour and tend to sort-out from the epithelium, whereas uniform increase in sal expression in its normal domain does not affect epithelial continuity. This function of Sal proteins was related to the regulation of the cell-membrane proteins Capricious and Tartan (Milán et al., 2002). In addition, we also identified FasIII as a membrane protein which expression depends on Sal in the central region of the wing blade. In this manner, it seems that Sal proteins regulate, by repression and activation, the transcription of several cell-adhesion proteins imposing an affinity marker that distinguish the central versus the peripheral regions of the wing blade (Cap and Tartan), and that also might contribute to cell adhesion (FasIII). Intriguingly, there are normal confrontations between sal expressing and not expressing cells in wild type discs, which are located in the anterior and posterior boundaries of the sal domain of expression. These confrontations are not associated to epithelial folds, indicating that wild type central (sal-expressing cells) and peripheral cells associate normally to each other. Nonetheless, some differences exist between these populations of cells, because the limits of sal expression correspond to weak clonal restrictions boundaries during normal development (Resino et al., 2002), indicating that the descendants of cells born in the sal domain of expression tend to remain in this domain. Therefore, it seems that the central and peripheral domains of the wing blade display some difference in affinity that is sufficient to promote a clonal restriction, but not able to trigger epithelial sorting during normal development. It is only when sal mutant cells, or cell expressing abnormally high levels of Sal, grow within the central domain that these differences became sufficient to trigger epithelial sorting. It has been argued that the central and peripheral domains of the wing blade have an intrinsic difference with respect to the response to Dpp signalling, and that this difference, of unknown nature, is set independently of Dpp signalling (Schwank et al., 2008). The distinct behaviour of the central domain with respect to abnormal confrontations between cells expressing different levels of Sal proteins agrees with this proposal, and suggests that Sal function contributes within this central territory to balance the expression of other undefined cell affinity markers involved in the differentiation of the central and peripheral territories.

A second role of Sal proteins is related to the promotion of cell proliferation within the central region of the wing blade. Thus, sal mutant clones are smaller than their twins, and when entire anterior or posterior compartments are formed by sal mutant cells their sizes are smaller than normal. We also found that mutant cells located in the central domain of the wing blade growing with lower than normal levels of Sal proteins tend to accumulate in the G2 phase of the cell cycle, and enter into mitosis at a lower rate than wild type cells. These mutant cells are also larger in size than wild type cells during pupal development. We speculate that Sal affects some aspect of the G2/M transition. The G2/M transition is regulated by String (stg), a phosphatase orthologous to cdc25 that de-phosphorylates the complex CycB/Cdk1. We observed that the expression of stg seems reduced in the central region of salm/salr mutant discs, but that forced stg expression does not rescue the proliferation defects of sal mutant cells, indicating that Sal proteins regulate other aspect than stg expression during G2/M progression. The Dpp pathway regulates growth by repressing brk expression, and it has been proposed that the Dpp/Brk system mostly limits proliferation in lateral areas of the wing disc (Schwank et al., 2008). We find that this aspect of Dpp/Brk function is independent of Sal activity, because clones of cells expressing Tkv^QD, which lack brk and have ectopic expression of Sal genes, still overgrow in peripheral regions of the disc in the absence of Sal function. In contrast Sal can rescue the proliferative defect of tkv mutant cells in the centre of the wing blade, suggesting that in this territory Sal mediates Dpp/Tkv activity. The analysis of tkv mutant cells suggested that they show a severe reduction in the number of cells in S and G2, and an increase in the number of cells in G1 (Martín-Castellanos and Edgar, 2002). This is in contrast to the behaviour of sal mutant cells, which tend to accumulate in the G2 phase of the cell cycle. To reconcile these apparently conflicting results, we suggests that the Dpp pathway regulates the cell cycle both at the G1/S and G2/M transitions, and does so using different mechanisms. In this scenario, Dpp/Tkv would promote G1/S independently of Sal, and G2/M in a Sal-dependent manner. In the absence of tkv, cells retaining some Sal expression would go through G2/M and accumulate at G1/S, and in the absence of Sal, cells would tend to accumulate at G2/M.

In addition to cell proliferation and cellular affinities, Sal function is also required for cell viability in the central region of the wing blade. Thus, the reduction in sal expression is associated to the activation of JNK signalling and to the induction of cell death, although these two events have a minor contribution in the generation of the sal mutant phenotype. In this manner, when cell death or JNK signalling is prevented in the absence of sal, the wings are still smaller than normal and display similar pattern defects as sal mutant wings. We don't know the mechanisms linking Sal with JNK signalling and cell death. It is well know that epithelial discontinuities in Tkv activity lead to JNK activation and cell death (Gibson and Perrimon, 2005), but in our hands JNK activity, as visualized by puc expression, occurs in situations where the appearance of the epithelium is entirely normal. Thus, wing discs in which sal expression is reduced in its entire domain show a normal epithelium with a robust induction of puc expression. In this situation, cell death is mostly observed in the periphery of the sal domain of expression.

The activities of Sal in regulating cell division, viability, epithelial integrity and vein positioning suggest that these proteins regulate most aspects of the patterning of the central domain of the wing blade, and they do so acting as downstream mediators of Dpp signalling. Salm and Salr belong to a family of transcription factors containing several Zn-fingers domains, and recent biochemical evidence suggests that they might be acting as transcriptional repressors (Sánchez et al., 2011). This is compatible with the function as a repressor of vertebrate Sall1, which interacts with histone deacetylase and other components of the chromatin remodelling complex NuRD (Lauberth et al., 2007). The understanding of the genetic hierarchy acting downstream of the Dpp/Brk/Sal system requires the identification of the target genes regulated by the Brk and Sal transcription factors, which likely would include cell cycle regulators, cell adhesion proteins and cytoskeleton components. We think that the identification and cellular characterization of Sal functional requirements constitutes a necessary step towards the identification of their target genes, and consequently to the reconstruction of the cascade of events triggered by Dpp signalling in the Drosophila wing.

We are very grateful to Ana López-Varea and Verónica Flores for their skilful technical help, and Antonio Baonza, Carlos Estella and Cristina Molnar for criticism that greatly improved the manuscript. We thank the Hybridome Bank at Iowa University, NIG in Japan, Bloomington Stock Center and other colleagues for providing the tools necessary for this work. Our work is supported by grants BFU2009-09403 and CSD-2007-00008 from the Spanish Ministerio de Economia y competitividad. The CBMSO enjoins the institutional support of the Ramón Areces foundation.